Autumn Banner Campaign 2017: Report
Goran S. Milovanovic, Data Analyst, WMDE
October, 2017
Feedback should be send to goran.milovanovic_ext@wikimedia.de.
The campaign is run from 2017/10/05 to 2017/10/13.
CURRENT UPDATE: Complete dataset, collected on 2017/10/14.
0. Preliminaries
0. 1 Data Acquisiton
NOTE: the Data Acquisition code chunk is not fully reproducible from this Report. The data are collected by running the script abc2017_PROD_OverallDailyUpdate.R on stat1005.eqiad.wmnet, collecting the data as .tsv and .csv files, copying manually, and processing locally. Run from stat1005 stat box by executing Rscript /home/goransm/RScripts/abc2017/abc2017_PROD_OverallDailyUpdate.R.
### --- Script: abc2017_PROD_OverallDailyUpdate.R
### --- the following runs on stat1005.eqiad.wmnet
### --- Rscript /home/goransm/RScripts/abc2017/abc2017_PROD_OverallDailyUpdate.R
### --- The script collects and wrangles all datasets
### --- for the WMDE Autumn Banner Campaign 2017.
### --- Goran S. Milovanovic, Data Analyst, WMDE
### --- September 26, 2017.
### -----------------------------------------------------------------------------
### 0. Setup
### -----------------------------------------------------------------------------
rm(list = ls())
library(dplyr)
library(tidyr)
library(stringr)
library(data.table)
startDate <- '2017-10-05'
endDate <- '2017-10-14'
bannerImpressionsDir <- '/home/goransm/_miscWMDE/abc2017_DataOUT/abc2017_OfficialDatasets/abc2017BannerImpressions/'
bannerClicksDir <- '/home/goransm/_miscWMDE/abc2017_DataOUT/abc2017_OfficialDatasets/abc2017BannerClicksLandingPages/'
dailyUpdateDir <- '/home/goransm/_miscWMDE/abc2017_DataOUT/abc2017_OfficialDatasets/abc2017_DailyUpdate/'
### -----------------------------------------------------------------------------
### 1. Banner Impressions
### -----------------------------------------------------------------------------
### --- Campaign Banner Tags:
# - (1) ?campaign=wmde_abc2017_bt1 - banner for Specific Task 1;
# - (2) ?campaign=wmde_abc2017_bt2 - banner for Specific Task 2;
# - (3) ?campaign=wmde_abc2017_bt3 - banner for Specific Task 3;
# - (4) ?campaign=wmde_abc2017_gib_lp - banner for the General Invitation
# - which leads to the Landing Page upon click;
# - (5) ?campaign=wmde_abc2017_gib_rg - banner for the General Invitation
# which leads directly to Registration upon click.
### --- HiveQL for everything from
### --- uri_host = 'de.wikipedia.org' and
### --- uri_path = '/beacon/impression'
### --- and then look up the desired tags.
### --- loop over date range, create query, fetch, and store
dateRange <- seq.POSIXt(from = as.POSIXlt(startDate, tz = "CET"),
to = as.POSIXlt(endDate, tz = "CET"),
by = 'hour')
dateRange <- dateRange[-length(dateRange)]
cetDateRange <- as.character(dateRange)
cetDateRange <- sapply(cetDateRange, function(x) {
strsplit(x, split = " ", fixed = T)[[1]][1]
})
names(dateRange) <- cetDateRange
dateRange <- as.POSIXlt(dateRange, tz = "UTC")
# - up to today:
today <- as.POSIXlt(Sys.time(), tz = "UTC")
w <- which(dateRange > today)
if (length(w) > 0) {
dateRange <- dateRange[-w]
}
dR <- list()
for (i in 1:length(dateRange)) {
dR[[i]] <- data.frame(
cetName = names(dateRange[i]),
utcYear = year(dateRange[i]),
utcMonth = month(dateRange[i]),
utcDay = mday(dateRange[i]),
utcHour = hour(dateRange[i])
)
}
dR <- rbindlist(dR)
dR <- dR %>%
group_by(cetName, utcYear, utcMonth, utcDay) %>%
summarise(utcHour = paste("hour=", utcHour, collapse = " OR ", sep = ""))
# - set outDir
outDir <- bannerImpressionsDir
setwd(outDir)
# - set HiveQL query dir:
for (i in 1:length(unique(dR$cetName))) {
wCetName <- which(dR$cetName %in% unique(dR$cetName)[i])
for (j in 1:length(wCetName)) {
# - construct HiveQL query:
y <- dR$utcYear[wCetName[j]]
m <- dR$utcMonth[wCetName[j]]
d <- dR$utcDay[wCetName[j]]
hour <- dR$utcHour[wCetName[j]]
q <- paste(
"USE wmf;
SELECT uri_query FROM webrequest
WHERE uri_host = 'de.wikipedia.org'
AND uri_path = '/beacon/impression'
AND year = ", y,
" AND month = ", m,
" AND day = ", d,
" AND (", hour, ");",
sep = "")
# - write hql
write(q, 'abc2017_BannerImpressions.hql')
# - prepare output file:
fileName <- "abc2017_BannerImpressions_"
fileName <- paste0(fileName,
as.character(unique(dR$cetName)[i]),
"_", j,
".tsv")
fileName <- paste0(outDir, fileName)
# - execute hql script:
hiveArgs <-
'beeline -f'
hiveInput <- paste0('abc2017_BannerImpressions.hql > ',
fileName)
# - command:
hiveCommand <- paste(hiveArgs, hiveInput)
system(command = hiveCommand, wait = TRUE)
}
}
### --- wrangle this dataSet
lF <- list.files()
lF <- lF[grepl(".tsv", lF, fixed = T)]
lF <- lF[grepl("Impressions", lF, fixed = T)]
### --- load Dataset:
# - count non-empty files:
c <- 0
dataSet <- list()
for (i in 1:length(lF)) {
dS <- readLines(lF[i], n = -1)
dS <- dS[8:(length(dS) - 1)]
if (length(dS) > 0) {
c <- c + 1
dS <- data.frame(query = dS,
date = strsplit(lF[i], split = "_", fixed = T)[[1]][3],
stringsAsFactors = F)
dataSet[[c]] <- dS
rm(dS); gc()
}
}
dataSet <- rbindlist(dataSet)
dataSet <- filter(dataSet,
grepl("WMDE_editor_campaign_autumn17",
query)
)
# - produce analytics dataset
banner <- str_extract(dataSet$query, "banner=(_|[[:alnum:]])+&")
banner <- gsub("banner=", "", banner, fixed = T)
banner <- gsub("&", "", banner, fixed = T)
impressionRate <- str_extract(dataSet$query, "recordImpressionSampleRate=([[:digit:]]|\\.)+&")
impressionRate <- gsub("recordImpressionSampleRate=", "", impressionRate, fixed = T)
impressionRate <- gsub("&", "", impressionRate, fixed = T)
impressionRate <- as.numeric(impressionRate)
status <- str_extract(dataSet$query, "status=([[:alnum:]]|[[:punct:]])+&")
status <- gsub("status=", "", status)
status <- gsub("&", "", status)
statusCode <- str_extract(dataSet$query, "statusCode=[[:digit:]]&")
statusCode <- gsub("statusCode=", "", statusCode)
statusCode <- gsub("&", "", statusCode)
campaignCategory <- str_extract(dataSet$query, "campaignCategory=[[:alnum:]]+&")
campaignCategory <- gsub("campaignCategory=", "", campaignCategory)
campaignCategory <- gsub("&", "", campaignCategory)
result <- str_extract(dataSet$query, "result=[[:alnum:]]+")
result <- gsub("result=", "", result)
result <- gsub("&", "", result)
qdate <- dataSet$date
# - as.data.frame()
dataSet <- data.frame(banner = banner,
impressionRate = impressionRate,
status = status,
statusCode = statusCode,
campaignCategory = campaignCategory,
result = result,
date = qdate,
stringsAsFactors = F)
# - store analytics dataset:
setwd(dailyUpdateDir)
dataSet <- dataSet[!is.na(dataSet$banner), ]
write.csv(dataSet, 'abc_BannerImpressions_update.csv')
### -----------------------------------------------------------------------------
### 2. Banner Clicks and Landing Page Views
### -----------------------------------------------------------------------------
### --- Landing/Registration pages:
# - Landing Page, Specific Tasks, Banners bt1, bt2, bt3
# - https://de.wikipedia.org/wiki/Wikipedia:Wikimedia_Deutschland/JetztMitmachen
# - Specific bt banner anchors:
# - bt1 - #Bebildern, bt2 - Aktualisieren, bt3 - #Belegen
# - Landing Page, General, Banner gib_lp
# - https://de.wikipedia.org/wiki/Wikipedia:Wikimedia_Deutschland/Mach_mit
# - Registration Page, banner gib_rg
# - https://de.wikipedia.org/wiki/Spezial:Benutzerkonto_anlegen
# - set outDir
outDir <- bannerClicksDir
setwd(outDir)
for (i in 1:length(unique(dR$cetName))) {
wCetName <- which(dR$cetName %in% unique(dR$cetName)[i])
for (j in 1:length(wCetName)) {
# - construct HiveQL query:
y <- dR$utcYear[wCetName[j]]
m <- dR$utcMonth[wCetName[j]]
d <- dR$utcDay[wCetName[j]]
hour <- dR$utcHour[wCetName[j]]
q <- paste(
"USE wmf;
SELECT uri_path, uri_query, referer FROM webrequest
WHERE uri_host = 'de.wikipedia.org'
AND (uri_path = '/wiki/Wikipedia:Wikimedia_Deutschland/JetztMitmachen' OR uri_path = '/wiki/Wikipedia:Wikimedia_Deutschland/Mach_mit' OR uri_path = '/wiki/Spezial:Benutzerkonto_anlegen')
AND year = ", y,
" AND month = ", m,
" AND day = ", d,
" AND (", hour, ");",
sep = "")
# - write hql
write(q, 'abc2017_BannerClicks.hql')
# - prepare output file:
fileName <- "abc2017_BannerClicks_"
fileName <- paste0(fileName,
as.character(unique(dR$cetName)[i]),
"_", j,
".tsv")
fileName <- paste0(outDir, fileName)
# - execute hql script:
hiveArgs <-
'beeline -f'
hiveInput <- paste0('abc2017_BannerClicks.hql > ',
fileName)
# - command:
hiveCommand <- paste(hiveArgs, hiveInput)
system(command = hiveCommand, wait = TRUE)
}
}
### --- Wrangle this dataset:
### --- Landing pages:
specTaskPage <- '/wiki/Wikipedia:Wikimedia_Deutschland/JetztMitmachen'
genInvPage <- '/wiki/Wikipedia:Wikimedia_Deutschland/Mach_mit'
regPage <- '/wiki/Spezial:Benutzerkonto_anlegen'
### --- Banner tags:
specTaskBanner1 <- '?campaign=wmde_abc2017_bt1'
specTaskBanner2 <- '?campaign=wmde_abc2017_bt2'
specTaskBanner3 <- '?campaign=wmde_abc2017_bt3'
genInvPage_rg <- '?campaign=wmde_abc2017_gib_rg'
genInvPage_lp <- '?campaign=wmde_abc2017_gib_lp'
### --- Dataset:
# - count non-empty files:
c <- 0
lF <- list.files()
lF <- lF[grepl('.tsv', lF, fixed = T)]
lF <- lF[grepl('Clicks', lF, fixed = T)]
dataSet <- list()
for (i in 1:length(lF)) {
dS <- readLines(lF[i], n = -1)
dS <- dS[8:(length(dS) - 2)]
if (length(dS > 0)) {
c <- c + 1
dS <- lapply(dS, function(x) {
dat <- strsplit(x, split = "\t", fixed = T)[[1]]
data.frame(page = dat[1], banner = dat[2], refer = dat[3], stringsAsFactors = F)
})
}
dS <- rbindlist(dS)
dS$date <- strsplit(lF[i], split = "_", fixed = T)[[1]][3]
dataSet[[c]] <- dS
rm(dS); gc()
}
dataSet <- rbindlist(dataSet)
# - replace values:
dataSet$page <- sapply(dataSet$page, function(x) {
strsplit(x, split = "/", fixed = T)[[1]][length(strsplit(x, split = "/", fixed = T)[[1]])]
})
dataSet$banner[which(dataSet$banner %in% specTaskBanner1)] <- 'BT1'
dataSet$banner[which(dataSet$banner %in% specTaskBanner2)] <- 'BT2'
dataSet$banner[which(dataSet$banner %in% specTaskBanner3)] <- 'BT3'
dataSet$banner[which(dataSet$banner %in% genInvPage_rg)] <- 'GIP_RG'
dataSet$banner[which(dataSet$banner %in% genInvPage_lp)] <- 'GIP_LP'
dataSet$banner <- paste(dataSet$banner, "_click", sep = "")
dataSet$banner[which(!(dataSet$banner %in% c('BT1_click',
'BT2_click',
'BT3_click',
'GIP_RG_click',
'GIP_LP_click')))] <- 'Other'
colnames(dataSet) <- c('Page', 'Source', 'Referer', 'Date')
### --- store abc_BannerClicksPageViews_Update.csv
write.csv(dataSet, file = "abc_BannerClicksPageViews_Non-Refined.csv")
dataSet$Source[dataSet$Page %in% 'Spezial:Benutzerkonto_anlegen' & dataSet$Source == 'Other'] <-
str_extract(dataSet$Referer[dataSet$Page %in% 'Spezial:Benutzerkonto_anlegen' & dataSet$Source == 'Other'],
"campaign=wmde_abc(.)+$")
dataSet$Source[dataSet$Page %in% 'Spezial:Benutzerkonto_anlegen' & grepl("wmde_abc2017_bt1", dataSet$Source)] <- "JetztMitmachen_BT1"
dataSet$Source[dataSet$Page %in% 'Spezial:Benutzerkonto_anlegen' & grepl("wmde_abc2017_bt2", dataSet$Source)] <- "JetztMitmachen_BT2"
dataSet$Source[dataSet$Page %in% 'Spezial:Benutzerkonto_anlegen' & grepl("wmde_abc2017_bt3", dataSet$Source)] <- "JetztMitmachen_BT3"
dataSet$Source[dataSet$Page %in% 'Spezial:Benutzerkonto_anlegen' & grepl("wmde_abc2017_gib_rg", dataSet$Source)] <- "GIP_RG_click"
dataSet$Source[dataSet$Page %in% 'Spezial:Benutzerkonto_anlegen' & grepl("wmde_abc2017_gib_lp", dataSet$Source)] <- "Mach_mit"
dataSet$Source[dataSet$Page %in% 'Spezial:Benutzerkonto_anlegen' & dataSet$Referer %in% '-'] <- "Unknown"
dataSet$Source[dataSet$Page %in% 'Mach_mit' & dataSet$Referer %in% '-'] <- "Unknown"
dataSet$Source[dataSet$Page %in% 'JetztMitmachen' & dataSet$Referer %in% '-'] <- "Unknown"
dataSet$Source[is.na(dataSet$Source)] <- 'Other'
dataSet$Referer <- NULL
### --- store abc_BannerClicksPageViews_Update.csv
setwd(dailyUpdateDir)
write.csv(dataSet, file = "abc_BannerClicksPageViews_Update.csv")
### -----------------------------------------------------------------------------
### 3. User Registration Data
### -----------------------------------------------------------------------------
# - NOTE: UTC timestamps - adjustment for CE(S)T introduced.
# - ServerSideAccountCreation_5487345
qCommand <- "mysql --defaults-file=/etc/mysql/conf.d/analytics-research-client.cnf -h analytics-store.eqiad.wmnet -A -e \"select * from log.ServerSideAccountCreation_5487345 where ((webHost = 'de.wikipedia.org') and (timestamp >= 20171004220000));\" > /home/goransm/_miscWMDE/abc2017_DataOUT/abc2017_OfficialDatasets/abc2017_DailyUpdate/abc2017_userRegistrations.tsv"
system(command = qCommand, wait = TRUE)
### -----------------------------------------------------------------------------
### 4. Guided Tour Data
### -----------------------------------------------------------------------------
# - NOTE: UTC timestamps - adjustment for CE(S)T introduced.
# - ServerSideAccountCreation_5487345
qCommand <- "mysql --defaults-file=/etc/mysql/conf.d/analytics-research-client.cnf -h analytics-store.eqiad.wmnet -A -e \"select * from log.GuidedTourExited_8690566 where ((webHost = 'de.wikipedia.org') and (timestamp >= 20171004220000));\" > /home/goransm/_miscWMDE/abc2017_DataOUT/abc2017_OfficialDatasets/abc2017_DailyUpdate/abc2017_guidedTours.tsv"
system(command = qCommand, wait = TRUE)
### -----------------------------------------------------------------------------
### 5. User Edits Data
### -----------------------------------------------------------------------------
# - get user IDs from registered:
lF <- list.files()
lF <- lF[grepl('userRegistrations', lF, fixed = T)]
userReg <- read.table(lF,
quote = "",
sep = "\t",
header = T,
check.names = F,
stringsAsFactors = F)
userReg <- userReg %>%
dplyr::select(event_userId, event_isSelfMade) %>%
filter(event_isSelfMade == 1)
# - uids:
uid <- userReg$event_userId
# - sql query
sqlQuery <- paste('SELECT COUNT(*) as edits, rev_user FROM revision WHERE rev_user IN (',
paste(uid, collapse = ", "),
') GROUP BY rev_user;',
sep = "")
mySqlCommand <- paste('mysql -h analytics-store.eqiad.wmnet dewiki -e ',
paste('"', sqlQuery, '" > ', sep = ""),
'/home/goransm/_miscWMDE/abc2017_DataOUT/abc2017_OfficialDatasets/abc2017_DailyUpdate/abc2017_userEdits.tsv', sep = "")
system(command = mySqlCommand,
wait = TRUE)0. 2 Abbreviations used in this Report
BT1Specific Task Banner Campaign 1 - “You can make Wikipedia more vivid! CTA: Learn how to add pictures to articles”BT2Specific Task Banner Campaign 2 - “You can improve the accuracy of Wikipedia! CTA: Learn how to improve articles”BT3Specific Task Banner Campaign 3 - “You can improve the reliability of Wikipedia! CTA: Learn how to add citations”GIBGeneral Invitation Banner - “Contribute to Wikipedia CTA: Create a user account”GIB_LPa version of the General Invitation Banner that leads to theMach_ mitlanding pageGIB_RGa version of the General Invitation Banner that leads directly to the registration pageMach_ mitthe landing page for the General Invitation BannerJetz_Mitmachen_the landing page for the Specific Task Banners
1. Campaign Banners and Pages
This section presents all data and statistics on the campaign banners and pages.
1.1 Banner Impressions
### --- extract only campaign relevant data
# - campaign banners:
BT1 <- 'WMDE_editor_campaign_autumn17_a'
BT2 <- 'WMDE_editor_campaign_autumn17_b'
BT3 <- 'WMDE_editor_campaign_autumn17_c'
GIB_RG <- 'WMDE_editor_campaign_autumn17_e'
GIB_LP <- 'WMDE_editor_campaign_autumn17_d'
dataSet <- read.csv('./_dailyUpdateDATA/abc_BannerImpressions_update.csv',
header = T,
row.names = 1,
check.names = F,
stringsAsFactors = F)
# - recode:
dataSet$banner <- recode(dataSet$banner,
'WMDE_editor_campaign_autumn17_a' = 'BT1',
'WMDE_editor_campaign_autumn17_b' = 'BT2',
'WMDE_editor_campaign_autumn17_c' = 'BT3',
'WMDE_editor_campaign_autumn17_e' = 'GIB_RG',
'WMDE_editor_campaign_autumn17_d' = 'GIB_LP',
'WMDE_editor_campaign_autumn17_a2' = 'BT1',
'WMDE_editor_campaign_autumn17_b2' = 'BT2',
'WMDE_editor_campaign_autumn17_c2' = 'BT3',
'WMDE_editor_campaign_autumn17_e2' = 'GIB_RG',
'WMDE_editor_campaign_autumn17_d2' = 'GIB_LP'
)
### --- Count daily banner impressions:
banImpSet <- dataSet %>%
filter(result %in% 'show') %>%
mutate(impressionRate = 1/impressionRate) %>%
group_by(banner, date) %>%
summarise(Count = sum(impressionRate))
colnames(banImpSet) <- c('Banner', 'Date', 'Count')
# - Campaign Chart Colors
campaignChartColors <- c('indianred1', 'indianred2', 'indianred3',
'cadetblue', 'cadetblue2')
names(campaignChartColors) <- c('BT1', 'BT2', 'BT3', 'GIB_LP', 'GIB_RG')
# - Visualize w. {ggplot2}
ggplot(banImpSet, aes(x = Date,
y = Count,
group = Banner,
color = Banner,
fill = Banner,
label = Count)) +
geom_path(size = .5) +
scale_fill_manual("legend", values = campaignChartColors) +
scale_color_manual("legend", values = campaignChartColors) +
scale_y_continuous(labels = comma) +
ggtitle('Autumn Banner Campaign 2017: Banner Impressions') +
theme_minimal() +
theme(axis.text.x = element_text(angle = 90, size = 8)) +
theme(plot.title = element_text(size = 10)) +
theme(legend.title = element_blank())
### --- Full Dataset (Table Report)
datatable(banImpSet)1.2.0 The Dataset
dataSet <- read.csv(paste('./_dailyUpdateDATA/', 'abc_BannerClicksPageViews_Update.csv', sep = ""),
header = T,
check.names = F,
row.names = 1,
stringsAsFactors = F) %>%
filter(Page %in% c('JetztMitmachen', 'Spezial:Benutzerkonto_anlegen', 'Mach_mit'))
# - fix 'GIP' -> 'GIB' in the dataset:
dataSet$Source <- gsub('GIP', 'GIB', dataSet$Source)
# - NOTE (TEMPORARY):
dataSet <- dataSet[1:(dim(dataSet)[1] - 2), ]
dataSet <- filter(dataSet,
!is.na(Page) & !is.na(Source) & !is.na(Date) & !(Source == "<NA>"))
# - Chart colorsgit
chartCols <- c('indianred1', 'indianred2', 'indianred3',
'cadetblue', 'cadetblue2',
'deepskyblue', 'violetred1', 'violetred2', 'violetred3',
'lightslategrey', 'lightgrey')
names(chartCols) <- c('BT1_click', 'BT2_click', 'BT3_click',
'GIB_LP_click', 'GIB_RG_click',
'Mach_mit', 'JetztMitmachen_BT1', 'JetztMitmachen_BT2', 'JetztMitmachen_BT3',
'Other', 'Unknown')
dataSet$Source <- factor(dataSet$Source, levels = c('BT1_click', 'BT2_click', 'BT3_click',
'GIB_LP_click', 'GIB_RG_click',
'Mach_mit', 'JetztMitmachen_BT1', 'JetztMitmachen_BT2', 'JetztMitmachen_BT3', 'Other', 'Unknown'))
# - Page Chart Colors
pageChartColors <- c('orange', 'deepskyblue', 'lightgreen')
names(pageChartColors) <- c('JetztMitmachen', 'Spezial:Benutzerkonto_anlegen', 'Mach_mit')
dataSet$Page <- factor(dataSet$Page,
levels = c('JetztMitmachen', 'Spezial:Benutzerkonto_anlegen', 'Mach_mit'))
# - Campaign Chart Colors
campaignChartColors <- c('indianred1', 'indianred2', 'indianred3',
'cadetblue', 'cadetblue2')
names(campaignChartColors) <- c('BT1', 'BT2', 'BT3', 'GIB_LP', 'GIB_RG')1.2.1A Landing Pages: Referers Overview
The following charts represents the breakdown of referers (i.e. sources) for the campaign pages: one registration page, and two landing pages.
### --- Banner clicks and Landing Page Views
# - Table Report
tableSet <- dataSet %>%
dplyr::group_by(Page, Source, Date) %>%
dplyr::summarise(Count = n()) %>%
dplyr::arrange(Date, Page, Source)
ggplot(tableSet, aes(x = Page,
y = Count,
group = Source,
color = Source,
fill = Source,
label = Count)) +
geom_bar(stat = "identity",
position = "dodge",
width = .35) +
scale_fill_manual("legend", values = chartCols) +
scale_color_manual("legend", values = chartCols) +
scale_y_continuous(labels = comma) +
ggtitle('Autumn Banner Campaign 2017:\nOverview of Landing Page Views sources') +
theme_minimal() +
theme(axis.text.x = element_text(angle = 90, size = 8, hjust = 1)) +
theme(plot.title = element_text(size = 10)) +
theme(legend.title = element_blank()) +
theme(panel.grid.major.x = element_blank()) +
theme(panel.grid.minor.x = element_blank()) +
theme(panel.background = element_blank())
1.2.2 Landing Pages: Referer Breakdown
The following three pie charts present a breakdown of referers (i.e. sources) for the Campaign pages (two landing pages and one registration page.
### --- Page Views: Sources
# - Spezial:Benutzerkonto_anlegen
pageSource <- dataSet %>%
dplyr::count(Page, Source) %>%
dplyr::group_by(Page) %>%
dplyr::mutate(Percent = n/sum(n))
pageSource$Percent <- paste(round(pageSource$Percent*100, 2), "%", sep = "")
pageSourcePlot <- filter(pageSource, Page %in% 'Spezial:Benutzerkonto_anlegen')
if (dim(pageSourcePlot)[1] > 0) {
ggplot(pageSourcePlot, aes(x = '',
y = n,
color = Source,
fill = Source,
label = Percent)) +
geom_bar(aes(x = '',
y = n,
color = Source,
fill = Source),
stat = "identity",
width = 1) +
coord_polar("y", start = 0) +
geom_text(aes(x = 1),
colour = "white",
fontface = "bold",
position = position_stack(vjust = 0.5),
size = 3,
show.legend = F) +
scale_fill_manual("legend", values = chartCols) +
scale_color_manual("legend", values = chartCols) +
scale_y_continuous(labels = comma) +
ggtitle('Autumn Banner Campaign 2017:\nPage Views Sources for Spezial:Benutzerkonto_anlegen') +
xlab("Outter = Count") + ylab("") +
theme_minimal() +
# theme(axis.text.x = element_blank()) +
theme(plot.title = element_text(size = 10)) +
theme(legend.title = element_blank()) +
theme(panel.grid.major.y = element_blank()) +
theme(panel.grid.minor.y = element_blank())
}
# - Spezial:Benutzerkonto_anlegen - Unknown/Other
pageSource <- dataSet %>%
filter(!(dataSet$Source %in% 'Other' | dataSet$Source %in% 'Unknown')) %>%
dplyr::count(Page, Source) %>%
dplyr::group_by(Page) %>%
dplyr::mutate(Percent = n/sum(n))
pageSource$Percent <- paste(round(pageSource$Percent*100, 2), "%", sep = "")
pageSourcePlot <- filter(pageSource, Page %in% 'Spezial:Benutzerkonto_anlegen')
if (dim(pageSourcePlot)[1] > 0) {
ggplot(pageSourcePlot, aes(x = '',
y = n,
color = Source,
fill = Source,
label = Percent)) +
geom_bar(aes(x = '',
y = n,
color = Source,
fill = Source),
stat = "identity",
width = 1) +
coord_polar("y", start = 0) +
geom_text(aes(x = 1),
colour = "white",
fontface = "bold",
position = position_stack(vjust = 0.5),
size = 3,
show.legend = F) +
scale_fill_manual("legend", values = chartCols) +
scale_color_manual("legend", values = chartCols) +
scale_y_continuous(labels = comma) +
ggtitle('Autumn Banner Campaign 2017:\nPage Views Sources for Spezial:Benutzerkonto_anlegen (Campaign only)') +
xlab("Outter = Count") + ylab("") +
theme_minimal() +
# theme(axis.text.x = element_blank()) +
theme(plot.title = element_text(size = 10)) +
theme(legend.title = element_blank()) +
theme(panel.grid.major.y = element_blank()) +
theme(panel.grid.minor.y = element_blank())
}
The following table presents the data in respect to the Campaign sources only:
### --- Full Dataset (Table Report)
datatable(pageSourcePlot)# - JetztMitmachen
pageSource <- dataSet %>%
dplyr::count(Page, Source) %>%
dplyr::group_by(Page) %>%
dplyr::mutate(Percent = n/sum(n))
pageSource$Percent <- paste(round(pageSource$Percent*100, 2), "%", sep = "")
pageSourcePlot <- filter(pageSource, Page %in% 'JetztMitmachen')
if (dim(pageSourcePlot)[1] > 0) {
ggplot(pageSourcePlot, aes(x = '',
y = n,
color = Source,
fill = Source,
label = Percent)) +
geom_bar(aes(x = '',
y = n,
color = Source,
fill = Source),
stat = "identity",
width = 1) +
coord_polar("y", start = 0) +
geom_text(aes(x = 1),
colour = "white",
fontface = "bold",
position = position_stack(vjust = 0.5),
size = 3,
show.legend = F) +
scale_fill_manual("legend", values = chartCols) +
scale_color_manual("legend", values = chartCols) +
scale_y_continuous(labels = comma) +
ggtitle('Autumn Banner Campaign 2017:\nPage Views Sources for JetztMitmachen') +
xlab("Outter = Count") + ylab("") +
theme_minimal() +
# theme(axis.text.x = element_blank()) +
theme(plot.title = element_text(size = 10)) +
theme(legend.title = element_blank()) +
theme(panel.grid.major.y = element_blank()) +
theme(panel.grid.minor.y = element_blank()) +
theme(panel.background = element_blank())
}
# - JetztMitmachen - minus Unknown/Other
pageSource <- dataSet %>%
filter(!(dataSet$Source %in% 'Other' | dataSet$Source %in% 'Unknown')) %>%
dplyr::count(Page, Source) %>%
dplyr::group_by(Page) %>%
dplyr::mutate(Percent = n/sum(n))
pageSource$Percent <- paste(round(pageSource$Percent*100, 2), "%", sep = "")
pageSourcePlot <- filter(pageSource, Page %in% 'JetztMitmachen')
if (dim(pageSourcePlot)[1] > 0) {
ggplot(pageSourcePlot, aes(x = '',
y = n,
color = Source,
fill = Source,
label = Percent)) +
geom_bar(aes(x = '',
y = n,
color = Source,
fill = Source),
stat = "identity",
width = 1) +
coord_polar("y", start = 0) +
geom_text(aes(x = 1),
colour = "white",
fontface = "bold",
position = position_stack(vjust = 0.5),
size = 3,
show.legend = F) +
scale_fill_manual("legend", values = chartCols) +
scale_color_manual("legend", values = chartCols) +
scale_y_continuous(labels = comma) +
ggtitle('Autumn Banner Campaign 2017:\nPage Views Sources for JetztMitmachen (Campaign only)') +
xlab("Outter = Count") + ylab("") +
theme_minimal() +
# theme(axis.text.x = element_blank()) +
theme(plot.title = element_text(size = 10)) +
theme(legend.title = element_blank()) +
theme(panel.grid.major.y = element_blank()) +
theme(panel.grid.minor.y = element_blank()) +
theme(panel.background = element_blank())
}
The following table presents the data in respect to the Campaign sources only:
### --- Full Dataset (Table Report)
datatable(pageSourcePlot)# - Mach_mit
pageSource <- dataSet %>%
dplyr::count(Page, Source) %>%
dplyr::group_by(Page) %>%
dplyr::mutate(Percent = n/sum(n))
pageSource$Percent <- paste(round(pageSource$Percent*100, 2), "%", sep = "")
pageSourcePlot <- filter(pageSource, Page %in% 'Mach_mit')
if (dim(pageSourcePlot)[1] > 0) {
ggplot(pageSourcePlot, aes(x = '',
y = n,
color = Source,
fill = Source,
label = Percent)) +
geom_bar(aes(x = '',
y = n,
color = Source,
fill = Source),
stat = "identity",
width = 1) +
coord_polar("y", start = 0) +
geom_text(aes(x = 1),
colour = "white",
fontface = "bold",
position = position_stack(vjust = 0.5),
size = 3,
show.legend = F) +
scale_fill_manual("legend", values = chartCols) +
scale_color_manual("legend", values = chartCols) +
scale_y_continuous(labels = comma) +
ggtitle('Autumn Banner Campaign 2017:\nPage Views Sources for Mach_mit') +
xlab("Outter = Count") + ylab("") +
theme_minimal() +
# theme(axis.text.x = element_blank()) +
theme(plot.title = element_text(size = 10)) +
theme(legend.title = element_blank()) +
theme(panel.grid.major.y = element_blank()) +
theme(panel.grid.minor.y = element_blank()) +
theme(panel.background = element_blank())
}
# - Mach_mit - minus Unknown/Other
pageSource <- dataSet %>%
filter(!(dataSet$Source %in% 'Other' | dataSet$Source %in% 'Unknown')) %>%
dplyr::count(Page, Source) %>%
dplyr::group_by(Page) %>%
dplyr::mutate(Percent = n/sum(n))
pageSource$Percent <- paste(round(pageSource$Percent*100, 2), "%", sep = "")
pageSourcePlot <- filter(pageSource, Page %in% 'Mach_mit')
if (dim(pageSourcePlot)[1] > 0) {
ggplot(pageSourcePlot, aes(x = '',
y = n,
color = Source,
fill = Source,
label = Percent)) +
geom_bar(aes(x = '',
y = n,
color = Source,
fill = Source),
stat = "identity",
width = 1) +
coord_polar("y", start = 0) +
geom_text(aes(x = 1),
colour = "white",
fontface = "bold",
position = position_stack(vjust = 0.5),
size = 3,
show.legend = F) +
scale_fill_manual("legend", values = chartCols) +
scale_color_manual("legend", values = chartCols) +
scale_y_continuous(labels = comma) +
ggtitle('Autumn Banner Campaign 2017:\nPage Views Sources for Mach_mit (Campaign only)') +
xlab("Outter = Count") + ylab("") +
theme_minimal() +
# theme(axis.text.x = element_blank()) +
theme(plot.title = element_text(size = 10)) +
theme(legend.title = element_blank()) +
theme(panel.grid.major.y = element_blank()) +
theme(panel.grid.minor.y = element_blank()) +
theme(panel.background = element_blank())
}
The following table presents the data in respect to the Campaign sources only:
### --- Full Dataset (Table Report)
datatable(pageSourcePlot)1.2.3 Banner Clicks: Campaign Total
The following charts represents the number of banner clicks for each campaign banner during the course of the campaign.
### --- Temporal Banner Clicks
# - Chart
clickPlotSet <- dataSet %>%
dplyr::select(Source, Date) %>%
dplyr::filter(grepl("_click", Source)) %>%
dplyr::group_by(Source, Date) %>%
dplyr::summarise(Count = n()) %>%
dplyr::arrange(Date)
ggplot(clickPlotSet, aes(x = Date,
y = Count,
group = Source,
color = Source,
fill = Source,
label = Count)) +
geom_path(size = .5) +
scale_fill_manual("legend", values = chartCols) +
scale_color_manual("legend", values = chartCols) +
scale_y_continuous(labels = comma) +
ggtitle('Autumn Banner Campaign 2017: Banner Clicks') +
theme_minimal() +
theme(axis.text.x = element_text(angle = 90, size = 8, hjust = 1)) +
theme(plot.title = element_text(size = 10)) +
theme(legend.title = element_blank()) +
theme(panel.background = element_blank())
### --- Full Dataset (Table Report)
datatable(clickPlotSet)1.2.4 Page Views: Campaign Total
The following charts presents (a) the number of page views for the two landing pages and one registration page during the course of the campaign, and then (b) encompassing only page views that were generated from the campaign.
### --- Temporal Page Views
pagePlotSet <- dataSet %>%
dplyr::select(Page, Date) %>%
dplyr::group_by(Page, Date) %>%
dplyr::summarise(Count = n()) %>%
dplyr::arrange(Date)
ggplot(pagePlotSet, aes(x = Date,
y = Count,
group = Page,
color = Page,
fill = Page,
label = Count)) +
geom_bar(stat = "identity",
position = "dodge",
width = .2) +
scale_fill_manual("legend", values = pageChartColors) +
scale_color_manual("legend", values = pageChartColors) +
scale_y_continuous(labels = comma) +
ggtitle('Autumn Banner Campaign 2017: Page Views') +
theme_minimal() +
theme(axis.text.x = element_text(angle = 90, size = 8, hjust = 1)) +
theme(plot.title = element_text(size = 10)) +
theme(legend.title = element_blank()) +
theme(panel.grid.major.x = element_blank()) +
theme(panel.grid.minor.x = element_blank()) +
theme(panel.background = element_blank())
### --- Temporal Page Views: Campaign only
pagePlotSet <- dataSet %>%
filter(!(dataSet$Source %in% 'Other' | dataSet$Source %in% 'Unknown')) %>%
dplyr::select(Page, Date) %>%
dplyr::group_by(Page, Date) %>%
dplyr::summarise(Count = n()) %>%
dplyr::arrange(Date)
ggplot(pagePlotSet, aes(x = Date,
y = Count,
group = Page,
color = Page,
fill = Page,
label = Count)) +
geom_bar(stat = "identity",
position = "dodge",
width = .2) +
scale_fill_manual("legend", values = pageChartColors) +
scale_color_manual("legend", values = pageChartColors) +
scale_y_continuous(labels = comma) +
ggtitle('Autumn Banner Campaign 2017: Page Views (Campaign only)') +
theme_minimal() +
theme(axis.text.x = element_text(angle = 90, size = 8, hjust = 1)) +
theme(plot.title = element_text(size = 10)) +
theme(legend.title = element_blank()) +
theme(panel.grid.major.x = element_blank()) +
theme(panel.grid.minor.x = element_blank()) +
theme(panel.background = element_blank())
The following table presents the data in respect to the Campaign sources only:
### --- Full Dataset (Table Report)
datatable(pagePlotSet)2. Campaign User Registrations
2. 0 Registrations
### --- Campaign User Registrations
lF <- list.files(path = "./_dailyUpdateDATA/")
lF <- lF[grepl('userRegistrations', lF, fixed = T)]
userReg <- read.table(paste("./_dailyUpdateDATA/", lF, sep = ""),
quote = "",
sep = "\t",
header = T,
check.names = F,
stringsAsFactors = F)
userReg$timestamp <- as.character(userReg$timestamp)
userReg$timestamp <- sapply(userReg$timestamp, function(x) {
y <- substr(x, 1, 4)
m <- substr(x, 5, 6)
d <- substr(x, 7, 8)
part1Date <- paste(y, m, d, sep = "-")
hr <- substr(x, 9, 10)
mi <- substr(x, 11, 12)
se <- substr(x, 13, 14)
part2Date <- paste(hr, mi, se, sep = ":")
paste(part1Date, part2Date, sep = " ")
})
userReg$timestamp <- as.POSIXct(userReg$timestamp, tz = "UTC")
timeDiff <-
as.POSIXct(as.character(Sys.time()), tz = "UTC") - as.POSIXct(as.character(Sys.time()), tz = "Europe/Berlin")
userReg$timestamp <- as.character(userReg$timestamp + timeDiff)
userReg$timestamp <- sapply(userReg$timestamp, function(x) {
y <- substr(x, 1, 4)
m <- substr(x, 6, 7)
d <- substr(x, 9, 10)
paste(y, m, d, sep = "-")
})
userReg <- userReg %>%
dplyr::select(id, event_userId, timestamp, event_isSelfMade, event_campaign) %>%
filter(event_isSelfMade == 1 & grepl("wmde_abc2017", event_campaign))
print(paste(dim(userReg)[1], " users have registered via the Campaign."))[1] "1054 users have registered via the Campaign."2. 1A User Registrations per Campaign (daily)
regPlotSet <- userReg %>%
group_by(event_campaign, timestamp) %>%
summarise(Registrations = n()) %>%
arrange(timestamp)
colnames(regPlotSet) <- c('Campaign', 'Date', 'Registrations')
regPlotSet$Campaign <- factor(toupper(gsub("wmde_abc2017_", "", regPlotSet$Campaign)))
ggplot(regPlotSet, aes(x = Date,
y = Registrations,
group = Campaign,
color = Campaign,
fill = Campaign)) +
geom_bar(stat = "identity",
position = "dodge",
width = .5) +
scale_fill_manual("legend", values = campaignChartColors) +
scale_color_manual("legend", values = campaignChartColors) +
scale_y_continuous(labels = comma) +
ggtitle('Autumn Banner Campaign 2017: User Registrations (daily)') +
theme_minimal() +
theme(axis.text.x = element_text(angle = 90, size = 8, hjust = 1)) +
theme(plot.title = element_text(size = 10)) +
theme(legend.title = element_blank()) +
theme(panel.grid.major.x = element_blank()) +
theme(panel.grid.minor.x = element_blank()) +
theme(panel.background = element_blank())
### --- Full Dataset (Table Report)
datatable(regPlotSet)2. 1B User Registrations per Campaign (totals)
regPlotSetTotal <- regPlotSet %>%
group_by(Campaign) %>%
summarise(Registrations = sum(Registrations)) %>%
arrange(Campaign)
ggplot(regPlotSetTotal, aes(x = Campaign,
y = Registrations,
group = Campaign,
color = Campaign,
fill = Campaign,
label = Registrations)) +
geom_bar(stat = "identity",
position = "dodge",
width = .5) +
geom_label(fill = "white", color = "black") +
scale_fill_manual("legend", values = campaignChartColors) +
scale_color_manual("legend", values = campaignChartColors) +
scale_y_continuous(labels = comma) +
ggtitle('Autumn Banner Campaign 2017: User Registrations (totals)') +
theme_minimal() +
theme(axis.text.x = element_text(angle = 90, size = 8, hjust = 1)) +
theme(plot.title = element_text(size = 10)) +
theme(legend.title = element_blank()) +
theme(panel.grid.major.x = element_blank()) +
theme(panel.grid.minor.x = element_blank()) +
theme(panel.background = element_blank())
2. 2 Total User Registrations daily
regPlotSetDaily <- userReg %>%
dplyr::filter(event_isSelfMade == 1 & grepl("wmde_abc2017", event_campaign)) %>%
group_by(timestamp) %>%
summarise(Registrations = n()) %>%
arrange(timestamp)
colnames(regPlotSetDaily) <- c('Date', 'Registrations')
ggplot(regPlotSetDaily, aes(x = Date,
y = Registrations,
label = Registrations)) +
geom_bar(stat = "identity",
position = "dodge",
width = .5,
fill = "darkblue",
color = "darkblue") +
geom_label() +
scale_y_continuous(labels = comma) +
ggtitle('Autumn Banner Campaign 2017: User Registrations') +
theme_minimal() +
theme(axis.text.x = element_text(angle = 90, size = 8, hjust = 1)) +
theme(plot.title = element_text(size = 10)) +
theme(legend.title = element_blank()) +
theme(panel.grid.major.x = element_blank()) +
theme(panel.grid.minor.x = element_blank()) +
theme(panel.background = element_blank())
### --- Full Dataset (Table Report)
datatable(regPlotSet)3. Campaign Guided Tour
3. 1A Guided Tour Point of Exit (daily)
### --- Campaign User Registrations
gTourData <- read.table("./_dailyUpdateDATA/abc2017_guidedTours.tsv",
quote = "",
sep = "\t",
header = T,
check.names = F,
stringsAsFactors = F)
# - clean up: gTourData
gTourData <- gTourData[which(!duplicated(gTourData$event_userId)), ]
gTourData <- gTourData[which(!(gTourData$event_userId == 0)), ]
gTourData <- gTourData[which(gTourData$event_userId %in% userReg$event_userId), ]
gTourData$timestamp <- as.character(gTourData$timestamp)
gTourData$timestamp <- sapply(gTourData$timestamp, function(x) {
y <- substr(x, 1, 4)
m <- substr(x, 5, 6)
d <- substr(x, 7, 8)
part1Date <- paste(y, m, d, sep = "-")
hr <- substr(x, 9, 10)
mi <- substr(x, 11, 12)
se <- substr(x, 13, 14)
part2Date <- paste(hr, mi, se, sep = ":")
paste(part1Date, part2Date, sep = " ")
})
gTourData$timestamp <- as.POSIXct(gTourData$timestamp, tz = "UTC")
timeDiff <-
as.POSIXct(as.character(Sys.time()), tz = "UTC") - as.POSIXct(as.character(Sys.time()), tz = "Europe/Berlin")
gTourData$timestamp <- as.character(gTourData$timestamp + timeDiff)
gTourData$timestamp <- sapply(gTourData$timestamp, function(x) {
y <- substr(x, 1, 4)
m <- substr(x, 6, 7)
d <- substr(x, 9, 10)
paste(y, m, d, sep = "-")
})
gTourData <- gTourData %>%
filter(event_tour %in% 'einfuhrung')
plotGTourData <- gTourData %>%
group_by(event_step, timestamp) %>%
summarise(Count = n())
colnames(plotGTourData) <- c('Tour Step', 'Date', 'Count')
ggplot(plotGTourData, aes(x = Date,
y = Count,
group = `Tour Step`,
color = `Tour Step`,
fill = `Tour Step`)) +
geom_bar(stat = "identity",
position = "dodge",
width = .5) +
scale_y_continuous(labels = comma) +
ggtitle('Autumn Banner Campaign 2017: Guided Tour Steps (daily)') +
theme_minimal() +
theme(axis.text.x = element_text(angle = 90, size = 8, hjust = 1)) +
theme(plot.title = element_text(size = 10)) +
theme(legend.title = element_blank()) +
theme(panel.grid.major.x = element_blank()) +
theme(panel.grid.minor.x = element_blank()) +
theme(panel.background = element_blank())
### --- Full Dataset (Table Report)
datatable(plotGTourData)3. 1B Guided Tour Point of Exit (totals)
### --- Campaign User Registrations
plotGTourDataTotal <- plotGTourData %>%
group_by(`Tour Step`) %>%
summarise(Count = sum(Count)) %>%
arrange(desc(Count))
plotGTourDataTotal$`Tour Step` <-
factor(plotGTourDataTotal$`Tour Step`,
levels = plotGTourDataTotal$`Tour Step`[order(plotGTourDataTotal$Count)])
ggplot(plotGTourDataTotal, aes(x = `Tour Step`,
y = Count,
group = `Tour Step`,
color = `Tour Step`,
fill = `Tour Step`,
label = Count)) +
geom_bar(stat = "identity",
position = "dodge",
width = .5) +
scale_y_continuous(labels = comma) +
geom_label(fill = "white", color = "black") +
ggtitle('Autumn Banner Campaign 2017: Guided Tour Steps (totals)') +
theme_minimal() +
theme(axis.text.x = element_text(angle = 90, size = 8, hjust = 1)) +
theme(plot.title = element_text(size = 10)) +
theme(legend.title = element_blank()) +
theme(panel.grid.major.x = element_blank()) +
theme(panel.grid.minor.x = element_blank()) +
theme(panel.background = element_blank()) +
theme(legend.position = 'None')
Number of users not exiting the Guided Tour:
nRegistered <- dim(userReg)[1]
nExitedGT <- dim(gTourData)[1]
print(paste(nRegistered - nExitedGT,
" users out of ",
nRegistered,
" (", round((nRegistered - nExitedGT)/nRegistered*100, 2), "%) did not exit the Campaign Guided Tour",
sep = ""))[1] "667 users out of 1054 (63.28%) did not exit the Campaign Guided Tour"3. 2 Exiting the Guided Tour at the Initial Step
How many users exit the Guided Tour at the initial step? NOTE: The Others category encompasses all users who did not exit at the initial step; they have either exited the Guided Tour later on or completed the tour.
exGTdata <- plotGTourData %>%
group_by(`Tour Step`) %>%
summarise(Count = sum(Count))
exGT1 <- exGTdata$Count[exGTdata$`Tour Step` %in% 'willkommen']
exGT2 <- nRegistered - exGT1
exGTourStep1 <- paste(exGT1, " (", round(exGT1/(exGT1 + exGT2)*100, 2), "%)", sep = "")
exGTourStep2 <- paste(exGT2, " (", round(exGT2/(exGT1 + exGT2)*100, 2), "%)", sep = "")
exGTour1 <- data.frame(`Users who exited at Step 1` = exGTourStep1,
`Others` = exGTourStep2,
check.names = F,
stringsAsFactors = F)
knitr::kable(exGTour1, format = "html") %>%
kable_styling(full_width = F, position = "left")| Users who exited at Step 1 | Others |
|---|---|
| 129 (12.24%) | 925 (87.76%) |
4. User Edits
4. 0 Proportion of Active Users
# - determine userIDs
userReg <- userReg %>%
dplyr::select(id, event_userId, timestamp, event_isSelfMade, event_campaign) %>%
filter(event_isSelfMade == 1 & grepl("wmde_abc2017", event_campaign))
userIDs <- userReg$event_userId
editData <- read.table("./_dailyUpdateDATA/abc2017_userEdits.tsv",
sep = "\t",
quote = "",
header = T,
check.names = F,
stringsAsFactors = F) %>%
filter(rev_user %in% userIDs)
plEditData <- editData %>%
group_by(edits) %>%
summarise(Count = n())
colnames(plEditData) <- c('Num.Edits', 'Count')
print(paste(sum(plEditData$Count),
" out of ",
dim(userReg)[1],
" registered users (",
round(sum(plEditData$Count)/dim(userReg)[1]*100, 2),
"%) have made at least one edit.",
sep = "")
)[1] "223 out of 1054 registered users (21.16%) have made at least one edit."4. 1 User Edits Distribution
The y-axis represents log(Number of users) to make the line plot more readable, while the data labels present exact user counts alongside the number of edits made.
ggplot(plEditData, aes(x = `Num.Edits`,
y = log(Count),
label = paste(Count, " (", `Num.Edits`, " edits)", sep = ""))
) +
geom_path(size = .25, color = "darkblue") +
geom_point(size = 1.5, color = "darkblue") +
geom_point(size = 1, color = "white") +
geom_text_repel(size = 3) +
scale_y_continuous(labels = comma) +
ylab('log(Num. of Users)') + xlab('Number of Edits') +
ggtitle('Autumn Banner Campaign 2017: User Edits Distribution') +
theme_minimal() +
theme(axis.text.x = element_text(angle = 0, size = 8, hjust = 1)) +
theme(plot.title = element_text(size = 10)) +
theme(legend.title = element_blank()) +
theme(panel.grid.major.x = element_blank()) +
theme(panel.grid.minor.x = element_blank()) +
theme(panel.background = element_blank())
4. 2 User Edits per Campaign
editCampaign <- left_join(editData, userReg,
by = c("rev_user" = "event_userId")) %>%
group_by(event_campaign) %>%
summarise(Edits = sum(edits))
colnames(editCampaign) <- c('Campaign', 'Edits')
# - recode:
editCampaign$Campaign <- recode(editCampaign$Campaign,
'wmde_abc2017_bt1' = 'BT1',
'wmde_abc2017_bt2' = 'BT2',
'wmde_abc2017_bt3' = 'BT3',
'wmde_abc2017_gib_rg' = 'GIB_RG',
'wmde_abc2017_gib_lp' = 'GIB_LP'
)
editCampaign$Campaign <- factor(editCampaign$Campaign,
levels = names(campaignChartColors))
ggplot(editCampaign, aes(x = Campaign,
y = Edits,
fill = Campaign,
color = Campaign,
label = Edits)) +
geom_bar(stat = "identity",
position = "dodge",
width = .5) +
geom_label(fill = "white", color = "black") +
scale_y_continuous(labels = comma) +
scale_fill_manual("legend", values = campaignChartColors) +
scale_color_manual("legend", values = campaignChartColors) +
ggtitle('Autumn Banner Campaign 2017: User Edits per Campaign') +
theme_minimal() +
theme(axis.text.x = element_text(angle = 0, size = 8, hjust = 1)) +
theme(plot.title = element_text(size = 10)) +
theme(legend.title = element_blank()) +
theme(panel.grid.major.x = element_blank()) +
theme(panel.grid.minor.x = element_blank()) +
theme(panel.background = element_blank())
### --- Full Dataset (Table Report)
datatable(editCampaign)4. 3 Percent of Active Users per Campaign
The percent of users who made any edits at all per campaign:
# - the dataset
editsMade <- left_join(userReg, editData,
by = c('event_userId' = 'rev_user'))
editsMade$event_campaign <- toupper(gsub("wmde_abc2017_", "", editsMade$event_campaign, fixed = T))
editsMade$edits[is.na(editsMade$edits)] <- 0
editsMade$Edit <- ifelse(editsMade$edits > 0, 'Edited', 'No edits')
editsMade <- dplyr::select(editsMade,
event_campaign, Edit)
colnames(editsMade)[1] <- 'Campaign'
editsMade <- editsMade %>%
group_by(Campaign, Edit) %>%
summarise(Count = n())
editsMade <- editsMade %>%
group_by(Campaign) %>%
mutate(Count = round(Count/sum(Count)*100, 2))
editsMade$Edit <- factor(editsMade$Edit, levels = c('Edited', 'No edits'))
ggplot(editsMade, aes(x = '', y = Count,
fill = Edit,
color = Edit,
group = Edit,
label = Count)) +
geom_bar(position = "stack",
stat = "identity",
width = 1,
color = "black") +
coord_polar("y", start = 0) +
facet_wrap(~ Campaign) +
scale_fill_manual("legend", values = c('firebrick', 'white')) +
scale_color_manual("legend", values = c('firebrick', 'white')) +
ggtitle('Autumn Banner Campaign 2017: User Edits Distributions per Campaign') +
xlab("") + ylab("Percent Edited") +
theme_minimal() +
theme(axis.text.x = element_text(angle = 0, size = 8, hjust = 1)) +
theme(plot.title = element_text(size = 10)) +
theme(legend.title = element_blank()) +
theme(panel.grid.major.x = element_blank()) +
theme(panel.grid.minor.x = element_blank()) +
theme(panel.background = element_blank()) 
4. 4 User Edits Breakdown
In the following table: No edits: users with zero edits, Edited: number of user who made any edits at all, 1 - 4 edits: number of users with 1 - 4 edits, 5 - 10 edits: number of users with 5 - 10 edits, and >10 edits: number of user with more than ten edits.
### --- Full Dataset (Table Report)
pltEdits <- as.tbl(editData) %>%
dplyr::group_by(edits) %>%
count()
edits0 <- dim(userReg)[1] - sum(plEditData$Count)
edits <- sum(pltEdits$n[pltEdits$edits > 0])
edits1_4 <- sum(pltEdits$n[pltEdits$edits >= 1 & pltEdits$edits <= 4])
edits5_10 <- sum(pltEdits$n[pltEdits$edits >= 5 & pltEdits$edits <= 10])
edits10 <- sum(pltEdits$n[pltEdits$edits > 10])
editClasses <- data.frame(`No edits` = edits0,
`Edited` = edits,
`1 - 4 edits` = edits1_4,
`5 - 10 edits` = edits5_10,
`> 10 edits` = edits10,
check.names = F,
stringsAsFactors = F)
knitr::kable(editClasses, format = "html") %>%
kable_styling(full_width = F, position = "left")| No edits | Edited | 1 - 4 edits | 5 - 10 edits | > 10 edits |
|---|---|---|---|---|
| 831 | 223 | 194 | 22 | 7 |
4. 5 User Edits and Guided Tour Exits
How many edits where made by users who did and did not exit the Campaign Guided Tour?
editGTData <- left_join(editData, userReg, by = c('rev_user' = 'event_userId'))
editGTData <- left_join(editGTData, gTourData, by = c('rev_user' = 'event_userId'))
exTourEdits <- sum(editGTData$edits[!is.na(editGTData$event_tour)])
notExTourEdits <- sum(editGTData$edits[is.na(editGTData$event_tour)])
exitedTourEdits <- paste(exTourEdits,
" (", round(exTourEdits/(exTourEdits + notExTourEdits)*100, 2), "%)",
sep = "")
notExitedTourEdits <- paste(notExTourEdits,
" (", round(notExTourEdits/(exTourEdits + notExTourEdits)*100, 2), "%)",
sep = "")
gtEdits <- data.frame(`Exited GT` = exitedTourEdits,
`Did not exit GT` = notExitedTourEdits,
check.names = F,
stringsAsFactors = F)
knitr::kable(gtEdits, format = "html") %>%
kable_styling(full_width = F, position = "left")| Exited GT | Did not exit GT |
|---|---|
| 292 (48.5%) | 310 (51.5%) |
4. 6 The Causal Effect of the Guided Tour Upon Editing
How does exiting vs. not exiting the Campaign Guided Tour influence whether the new user will make at least one edit or not? The following contingency table presents the number of registered users who made any edits at all (vs. those did not edit) separately for those who did and did not exit the Guided Tour.
userRegGT <- left_join(userReg, gTourData,
by = 'event_userId')
userRegGT <- left_join(userRegGT, editData,
by = c('event_userId' = 'rev_user'))
# - Contingency Table:
a <- length(userRegGT$event_userId[!is.na(userRegGT$edits) & is.na(userRegGT$event_tour)])
b <- length(userRegGT$event_userId[is.na(userRegGT$edits) & is.na(userRegGT$event_tour)])
c <- length(userRegGT$event_userId[!is.na(userRegGT$edits) & !is.na(userRegGT$event_tour)])
d <- length(userRegGT$event_userId[is.na(userRegGT$edits) & !is.na(userRegGT$event_tour)])
ct <- data.frame(`Edited` = c(a, c),
`No edits` = c(b, d),
check.names = F)
rownames(ct) <- c('GT Completed', 'GT Exited')
# - deltaP:
deltaP <- a/(a+b) - c/(c+d)
if (deltaP >= 0) {
causalP <- deltaP/(1 - c/(c+d))
} else {
causalP <- -deltaP/(c/(c+d))
}
knitr::kable(ct, format = "html") %>%
kable_styling(full_width = F, position = "left")| Edited | No edits | |
|---|---|---|
| GT Completed | 103 | 564 |
| GT Exited | 120 | 267 |
The estimate of the Causal Power (it can range from 0 = no causal influence at all, to 1 = a cause completelly sufficient to bring about its effect) of the Guided Tour to bring about any edits at all is:
ceffect <- ifelse(deltaP >= 0, 'generative effect', 'preventive effect')
print(paste('Guided Tour Causal Power: ', round(causalP, 2), ' (', ceffect, ')', sep = ""))[1] "Guided Tour Causal Power: 0.5 (preventive effect)"print(paste('(NOTE: with a value of a probabilistic contrast deltaP of): ', round(deltaP, 2), sep = ""))[1] "(NOTE: with a value of a probabilistic contrast deltaP of): -0.16"SUGGESTION: remove the Guided Tour from our future campaigns; it has an preventive effect upon the number of new user edits.
4. 7 Guided Tour and the number of user edits
How does exiting vs. not exiting the Campaign Guided Tour influence how many edits will a new user make?
userRegGT1 <- filter(userRegGT, !is.na(edits))
userRegGT1$event_tour <- ifelse(is.na(userRegGT1$event_tour), "Completed", "Exited")
dataGTEdits <- dplyr::select(userRegGT1, event_tour, edits) %>%
group_by(event_tour) %>%
summarise(`Num.edits` = sum(edits))
knitr::kable(dataGTEdits, format = "html") %>%
kable_styling(full_width = F, position = "left")| event_tour | Num.edits |
|---|---|
| Completed | 310 |
| Exited | 292 |
In conclusion, those new users who had completed the Guided Tour have also made a slightly higher number of edits.
5. Campaign Evaluation
5.1 A/B Testing: Campaign Banners
5.1A User Registrations
Prepare priors and data.
regData <- regPlotSet %>%
group_by(Campaign) %>%
summarise(Registrations = sum(Registrations))
viewData <- clickPlotSet %>%
group_by(Source) %>%
summarise(Clicks = sum(Count))
viewData$Source <- gsub("_click", "", viewData$Source)
regData <- left_join(regData, viewData, by = c('Campaign' = 'Source'))# - Uninformative prior:
priorAlpha <- 1
priorBeta <- 1
# - Data:
BT1Data <- c(rep(1, regData$Registrations[1]), rep(0, regData$Clicks[1] - regData$Registrations[1]))
BT2Data <- c(rep(1, regData$Registrations[2]), rep(0, regData$Clicks[2] - regData$Registrations[2]))
BT3Data <- c(rep(1, regData$Registrations[3]), rep(0, regData$Clicks[3] - regData$Registrations[3]))
GIB_LPData <- c(rep(1, regData$Registrations[4]), rep(0, regData$Clicks[4] - regData$Registrations[4]))
GIB_RGData <- c(rep(1, regData$Registrations[5]), rep(0, regData$Clicks[5] - regData$Registrations[5]))
# - Posteriors:
postB1Alpha <- priorAlpha + sum(BT1Data)
postB1Beta <- priorBeta + length(BT1Data) - sum(BT1Data)
postB2Alpha <- priorAlpha + sum(BT2Data)
postB2Beta <- priorBeta + length(BT2Data) - sum(BT2Data)
postB3Alpha <- priorAlpha + sum(BT3Data)
postB3Beta <- priorBeta + length(BT3Data) - sum(BT3Data)
postGIB_LPAlpha <- priorAlpha + sum(GIB_LPData)
postGIB_LPBeta <- priorBeta + length(GIB_LPData) - sum(GIB_LPData)
postGIB_RGAlpha <- priorAlpha + sum(GIB_RGData)
postGIB_RGBeta <- priorBeta + length(GIB_RGData) - sum(GIB_RGData)
# - Number of Monte Carlo samples:
mcN <- 1e6Summary
The GIB_RG banner dominates all other in terms of the probability of user registration. As of the BT campaigns: BT2 performs relativelly better than BT1 and BT2 which do not differ (or differ only slightly) between each other. The most important findings are:
- the dominance of the
GIBbanners over theBTbanners, - the dominance of
GIB_RGover any other banner.
NOTE. I use the term campaign lift in the following sections to refer to a difference in the probability of user registration in respect to any pair of banners that were used during the ABC2017 campaign. Theoretically, a campaign lift would be the difference (in some KPI) between a group of users who were exposed to the campaign and the control group (no exposure). If we would assess the ABC2017 campaign in this manner then it would be natural to take GIB_RG as a control, and compare all other groups (BT1, BT2, BT3, and GIB_LP) to it; in that case, we would observe that ABC2017 campaign had no lift in respect to the control group in terms of user registrations.
The following sections provide the results of pairwise Bayesian A/B tests across the campaign banners. TECHNICAL NOTE. Uniform Beta(1, 1) priors (assuming no prior knowlegde on the probability of a banner click leading to a registration) and 1,000,000 Monte Carlo samples from the posteriors were used.
Evaluation: BT1 vs. BT2
BT1Samples <- rbeta(mcN, postB1Alpha, postB1Beta)
BT2Samples <- rbeta(mcN, postB2Alpha, postB2Beta)
t_percentDiff <- (mean(BT1Samples) - mean(BT2Samples))/mean(BT2Samples)*100
pBT1_BT2 <- mean((BT1Samples > BT2Samples))
# - Probability of v1 better than v2:
print(paste('The probability of BT1 having more user registrations than BT2 is: ', pBT1_BT2))[1] "The probability of BT1 having more user registrations than BT2 is: 0"# - v1 Campaign Lift
percentDiff <- (BT1Samples - BT2Samples)/BT2Samples*100
percentDiff <- data.frame(percentDiff = percentDiff,
area = ifelse(percentDiff <= 0, '<= 0', '> 0'),
stringsAsFactors = T)
print(paste('The percent lift that BT1 has over BT2 (',
round(t_percentDiff, 2),
'%) lies in the interval (',
as.character(round(quantile(percentDiff$percentDiff, .025)), 2),
'%, ',
as.character(round(quantile(percentDiff$percentDiff, .975)), 2),
'%) with 95% certainty.',
sep = ""))[1] "The percent lift that BT1 has over BT2 (-54.62%) lies in the interval (-63%, -44%) with 95% certainty."ggplot(percentDiff, aes(x = percentDiff,
fill = area)) +
geom_histogram(binwidth = .1, alpha = .5) +
scale_y_continuous(labels = comma) +
xlab('(BT1 - BT2)/BT1') + ylab('Density') +
ggtitle('BT1/BT2 Campaign Lift') +
theme_minimal() +
theme(plot.title = element_text(size = 10))
Evaluation: BT1 vs. BT3
mcN <- 1e5
BT1Samples <- rbeta(mcN, postB1Alpha, postB1Beta)
BT3Samples <- rbeta(mcN, postB3Alpha, postB3Beta)
t_percentDiff <- (mean(BT1Samples) - mean(BT3Samples))/mean(BT3Samples)*100
pBT1_BT3 <- mean((BT1Samples > BT3Samples))
# - Probability of v1 better than v2:
print(paste('The probability of BT1 having more user registrations than BT3 is: ', pBT1_BT3))[1] "The probability of BT1 having more user registrations than BT3 is: 0.3196"# - v1 Campaign Lift
percentDiff <- (BT1Samples - BT3Samples)/BT3Samples*100
percentDiff <- data.frame(percentDiff = percentDiff,
area = ifelse(percentDiff <= 0, '<= 0', '> 0'),
stringsAsFactors = T)
print(paste('The percent lift that BT1 has over BT3 (',
round(t_percentDiff, 2),
'%) lies in the interval (',
as.character(round(quantile(percentDiff$percentDiff, .025)), 2),
'%, ',
as.character(round(quantile(percentDiff$percentDiff, .975)), 2),
'%) with 95% certainty.',
sep = ""))[1] "The percent lift that BT1 has over BT3 (-5.69%) lies in the interval (-26%, 20%) with 95% certainty."ggplot(percentDiff, aes(x = percentDiff,
fill = area)) +
geom_histogram(binwidth = .1, alpha = .5) +
scale_y_continuous(labels = comma) +
xlab('(BT1 - BT3)/BT3') + ylab('Density') +
ggtitle('BT1/BT3 Campaign Lift') +
theme_minimal() +
theme(plot.title = element_text(size = 10))
Evaluation: BT1 vs. GIB_LP
mcN <- 1e5
BT1Samples <- rbeta(mcN, postB1Alpha, postB1Beta)
GIB_LPSamples <- rbeta(mcN, postGIB_LPAlpha, postGIB_LPBeta)
t_percentDiff <- (mean(BT1Samples) - mean(GIB_LPSamples))/mean(GIB_LPSamples)*100
pBT1_GIB_LP <- mean((BT1Samples > GIB_LPSamples))
# - Probability of v1 better than v2:
print(paste('The probability of BT1 having more user registrations than GIB_LP is: ', pBT1_GIB_LP))[1] "The probability of BT1 having more user registrations than GIB_LP is: 0"# - v1 Campaign Lift
percentDiff <- (BT1Samples - GIB_LPSamples)/GIB_LPSamples*100
percentDiff <- data.frame(percentDiff = percentDiff,
area = ifelse(percentDiff <= 0, '<= 0', '> 0'),
stringsAsFactors = T)
print(paste('The percent lift that BT1 has over GIB_LP (',
round(t_percentDiff, 2),
'%) lies in the interval (',
as.character(round(quantile(percentDiff$percentDiff, .025)), 2),
'%, ',
as.character(round(quantile(percentDiff$percentDiff, .975)), 2),
'%) with 95% certainty.',
sep = ""))[1] "The percent lift that BT1 has over GIB_LP (-66.78%) lies in the interval (-74%, -58%) with 95% certainty."ggplot(percentDiff, aes(x = percentDiff,
fill = area)) +
geom_histogram(binwidth = .1, alpha = .5) +
scale_y_continuous(labels = comma) +
xlab('(BT1 - GIB_LP)/GIB_LP') + ylab('Density') +
ggtitle('BT1/GIB_LP Campaign Lift') +
theme_minimal() +
theme(plot.title = element_text(size = 10))
Evaluation: BT1 vs. GIB_RG
mcN <- 1e5
BT1Samples <- rbeta(mcN, postB1Alpha, postB1Beta)
GIB_RGSamples <- rbeta(mcN, postGIB_LPAlpha, postGIB_RGBeta)
t_percentDiff <- (mean(BT1Samples) - mean(GIB_RGSamples))/mean(GIB_RGSamples)*100
pBT1_GIB_RG <- mean((BT1Samples > GIB_RGSamples))
# - Probability of v1 better than v2:
print(paste('The probability of BT1 having more user registrations than GIB_RG is: ', pBT1_GIB_RG))[1] "The probability of BT1 having more user registrations than GIB_RG is: 0"# - v1 Campaign Lift
percentDiff <- (BT1Samples - GIB_RGSamples)/GIB_RGSamples*100
percentDiff <- data.frame(percentDiff = percentDiff,
area = ifelse(percentDiff <= 0, '<= 0', '> 0'),
stringsAsFactors = T)
print(paste('The percent lift that BT1 has over GIB_RG (',
round(t_percentDiff, 2),
'%) lies in the interval (',
as.character(round(quantile(percentDiff$percentDiff, .025)), 2),
'%, ',
as.character(round(quantile(percentDiff$percentDiff, .975)), 2),
'%) with 95% certainty.',
sep = ""))[1] "The percent lift that BT1 has over GIB_RG (-66.33%) lies in the interval (-73%, -58%) with 95% certainty."ggplot(percentDiff, aes(x = percentDiff,
fill = area)) +
geom_histogram(binwidth = .1, alpha = .5) +
scale_y_continuous(labels = comma) +
xlab('(BT1 - GIB_RG)/GIB_RG') + ylab('Density') +
ggtitle('BT1/GIB_RG Campaign Lift') +
theme_minimal() +
theme(plot.title = element_text(size = 10))
Evaluation: BT2 vs. BT3
mcN <- 1e5
BT2Samples <- rbeta(mcN, postB2Alpha, postB2Beta)
BT3Samples <- rbeta(mcN, postB3Alpha, postB3Beta)
t_percentDiff <- (mean(BT2Samples) - mean(BT3Samples))/mean(BT3Samples)*100
pBT2_BT3 <- mean((BT2Samples > BT3Samples))
# - Probability of v1 better than v2:
print(paste('The probability of BT2 having more user registrations than BT3 is: ', pBT2_BT3))[1] "The probability of BT2 having more user registrations than BT3 is: 1"# - v1 Campaign Lift
percentDiff <- (BT2Samples - BT3Samples)/BT3Samples*100
percentDiff <- data.frame(percentDiff = percentDiff,
area = ifelse(percentDiff <= 0, '<= 0', '> 0'),
stringsAsFactors = T)
print(paste('The percent lift that BT2 has over BT3 (',
round(t_percentDiff, 2),
'%) lies in the interval (',
as.character(round(quantile(percentDiff$percentDiff, .025)), 2),
'%, ',
as.character(round(quantile(percentDiff$percentDiff, .975)), 2),
'%) with 95% certainty.',
sep = ""))[1] "The percent lift that BT2 has over BT3 (107.7%) lies in the interval (69%, 158%) with 95% certainty."ggplot(percentDiff, aes(x = percentDiff,
fill = area)) +
geom_histogram(binwidth = .1, alpha = .5) +
scale_y_continuous(labels = comma) +
xlab('(BT2 - BT3)/BT3') + ylab('Density') +
ggtitle('BT2/BT3 Campaign Lift') +
theme_minimal() +
theme(plot.title = element_text(size = 10))
Evaluation: BT2 vs. GIB_LP
mcN <- 1e5
BT2Samples <- rbeta(mcN, postB2Alpha, postB2Beta)
GIB_LPSamples <- rbeta(mcN, postGIB_LPAlpha, postGIB_LPBeta)
t_percentDiff <- (mean(BT2Samples) - mean(GIB_LPSamples))/mean(GIB_LPSamples)*100
pBT2_GIB_LP <- mean((BT2Samples > GIB_LPSamples))
# - Probability of v1 better than v2:
print(paste('The probability of BT2 having more user registrations than GIB_LP is: ', pBT2_GIB_LP))[1] "The probability of BT2 having more user registrations than GIB_LP is: 7e-04"# - v1 Campaign Lift
percentDiff <- (BT2Samples - GIB_LPSamples)/GIB_LPSamples*100
percentDiff <- data.frame(percentDiff = percentDiff,
area = ifelse(percentDiff <= 0, '<= 0', '> 0'),
stringsAsFactors = T)
print(paste('The percent lift that BT2 has over GIB_LP (',
round(t_percentDiff, 2),
'%) lies in the interval (',
as.character(round(quantile(percentDiff$percentDiff, .025)), 2),
'%, ',
as.character(round(quantile(percentDiff$percentDiff, .975)), 2),
'%) with 95% certainty.',
sep = ""))[1] "The percent lift that BT2 has over GIB_LP (-26.79%) lies in the interval (-39%, -11%) with 95% certainty."ggplot(percentDiff, aes(x = percentDiff,
fill = area)) +
geom_histogram(binwidth = .1, alpha = .5) +
scale_y_continuous(labels = comma) +
xlab('(BT2 - GIB_LP)/GIB_LP') + ylab('Density') +
ggtitle('BT2/GIB_LP Campaign Lift') +
theme_minimal() +
theme(plot.title = element_text(size = 10))
Evaluation: BT2 vs. GIB_RG
mcN <- 1e5
BT2Samples <- rbeta(mcN, postB2Alpha, postB2Beta)
GIB_RGSamples <- rbeta(mcN, postGIB_RGAlpha, postGIB_RGBeta)
t_percentDiff <- (mean(BT2Samples) - mean(GIB_RGSamples))/mean(GIB_RGSamples)*100
pBT2_GIB_RG <- mean((BT2Samples > GIB_RGSamples))
# - Probability of v1 better than v2:
print(paste('The probability of BT2 having more user registrations than GIB_RG is: ', pBT2_GIB_RG))[1] "The probability of BT2 having more user registrations than GIB_RG is: 0"# - v1 Campaign Lift
percentDiff <- (BT2Samples - GIB_RGSamples)/GIB_RGSamples*100
percentDiff <- data.frame(percentDiff = percentDiff,
area = ifelse(percentDiff <= 0, '<= 0', '> 0'),
stringsAsFactors = T)
print(paste('The percent lift that BT2 has over GIB_RG (',
round(t_percentDiff, 2),
'%) lies in the interval (',
as.character(round(quantile(percentDiff$percentDiff, .025)), 2),
'%, ',
as.character(round(quantile(percentDiff$percentDiff, .975)), 2),
'%) with 95% certainty.',
sep = ""))[1] "The percent lift that BT2 has over GIB_RG (-62.91%) lies in the interval (-68%, -57%) with 95% certainty."ggplot(percentDiff, aes(x = percentDiff,
fill = area)) +
geom_histogram(binwidth = .1, alpha = .5) +
scale_y_continuous(labels = comma) +
xlab('(BT2 - GIB_RG)/GIB_RG') + ylab('Density') +
ggtitle('BT2/GIB_RG Campaign Lift') +
theme_minimal() +
theme(plot.title = element_text(size = 10))
Evaluation: BT3 vs. GIB_LP
mcN <- 1e5
BT3Samples <- rbeta(mcN, postB3Alpha, postB3Beta)
GIB_LPSamples <- rbeta(mcN, postGIB_LPAlpha, postGIB_LPBeta)
t_percentDiff <- (mean(BT3Samples) - mean(GIB_LPSamples))/mean(GIB_LPSamples)*100
pBT3_GIB_LP <- mean((BT3Samples > GIB_LPSamples))
# - Probability of v1 better than v2:
print(paste('The probability of BT3 having more user registrations than GIB_LP is: ', pBT3_GIB_LP))[1] "The probability of BT3 having more user registrations than GIB_LP is: 0"# - v1 Campaign Lift
percentDiff <- (BT3Samples - GIB_LPSamples)/GIB_LPSamples*100
percentDiff <- data.frame(percentDiff = percentDiff,
area = ifelse(percentDiff <= 0, '<= 0', '> 0'),
stringsAsFactors = T)
print(paste('The percent lift that BT3 has over GIB_LP (',
round(t_percentDiff, 2),
'%) lies in the interval (',
as.character(round(quantile(percentDiff$percentDiff, .025)), 2),
'%, ',
as.character(round(quantile(percentDiff$percentDiff, .975)), 2),
'%) with 95% certainty.',
sep = ""))[1] "The percent lift that BT3 has over GIB_LP (-64.79%) lies in the interval (-72%, -56%) with 95% certainty."ggplot(percentDiff, aes(x = percentDiff,
fill = area)) +
geom_histogram(binwidth = .1, alpha = .5) +
scale_y_continuous(labels = comma) +
xlab('(BT3 - GIB_LP)/GIB_LP') + ylab('Density') +
ggtitle('BT3/GIB_LP Campaign Lift') +
theme_minimal() +
theme(plot.title = element_text(size = 10))
Evaluation: BT3 vs. GIB_RG
mcN <- 1e5
BT3Samples <- rbeta(mcN, postB2Alpha, postB2Beta)
GIB_RGSamples <- rbeta(mcN, postGIB_RGAlpha, postGIB_RGBeta)
t_percentDiff <- (mean(BT3Samples) - mean(GIB_RGSamples))/mean(GIB_RGSamples)*100
pBT3_GIB_RG <- mean((BT3Samples > GIB_RGSamples))
# - Probability of v1 better than v2:
print(paste('The probability of BT3 having more user registrations than GIB_RG is: ', pBT3_GIB_RG))[1] "The probability of BT3 having more user registrations than GIB_RG is: 0"# - v1 Campaign Lift
percentDiff <- (BT3Samples - GIB_RGSamples)/GIB_RGSamples*100
percentDiff <- data.frame(percentDiff = percentDiff,
area = ifelse(percentDiff <= 0, '<= 0', '> 0'),
stringsAsFactors = T)
print(paste('The percent lift that BT3 has over GIB_RG (',
round(t_percentDiff, 2),
'%) lies in the interval (',
as.character(round(quantile(percentDiff$percentDiff, .025)), 2),
'%, ',
as.character(round(quantile(percentDiff$percentDiff, .975)), 2),
'%) with 95% certainty.',
sep = ""))[1] "The percent lift that BT3 has over GIB_RG (-62.92%) lies in the interval (-68%, -57%) with 95% certainty."ggplot(percentDiff, aes(x = percentDiff,
fill = area)) +
geom_histogram(binwidth = .1, alpha = .5) +
scale_y_continuous(labels = comma) +
xlab('(BT3 - GIB_RG)/GIB_RG') + ylab('Density') +
ggtitle('BT3/GIB_RG Campaign Lift') +
theme_minimal() +
theme(plot.title = element_text(size = 10))
Evaluation: GIB_LP vs. GIB_RG
mcN <- 1e5
GIB_LPSamples <- rbeta(mcN, postB2Alpha, postB2Beta)
GIB_RGSamples <- rbeta(mcN, postGIB_RGAlpha, postGIB_RGBeta)
t_percentDiff <- (mean(GIB_LPSamples) - mean(GIB_RGSamples))/mean(GIB_RGSamples)*100
pGIB_LP_GIB_RG <- mean((GIB_LPSamples > GIB_RGSamples))
# - Probability of v1 better than v2:
print(paste('The probability of GIB_LP having more user registrations than GIB_RG is: ', pGIB_LP_GIB_RG))[1] "The probability of GIB_LP having more user registrations than GIB_RG is: 0"# - v1 Campaign Lift
percentDiff <- (GIB_LPSamples - GIB_RGSamples)/GIB_RGSamples*100
percentDiff <- data.frame(percentDiff = percentDiff,
area = ifelse(percentDiff <= 0, '<= 0', '> 0'),
stringsAsFactors = T)
print(paste('The percent lift that GIB_LP has over GIB_RG (',
round(t_percentDiff, 2),
'%) lies in the interval (',
as.character(round(quantile(percentDiff$percentDiff, .025)), 2),
'%, ',
as.character(round(quantile(percentDiff$percentDiff, .975)), 2),
'%) with 95% certainty.',
sep = ""))[1] "The percent lift that GIB_LP has over GIB_RG (-62.91%) lies in the interval (-68%, -57%) with 95% certainty."ggplot(percentDiff, aes(x = percentDiff,
fill = area)) +
geom_histogram(binwidth = .1, alpha = .5) +
scale_y_continuous(labels = comma) +
xlab('(GIB_LP - GIB_RG)/GIB_RG') + ylab('Density') +
ggtitle('GIB_LP/GIB_RG Campaign Lift') +
theme_minimal() +
theme(plot.title = element_text(size = 10))
5.1B User Edits
Summary
What we know almost certainly is that, in general, the BT banners dominate the GIB_LP and GIB_RG banners in terms of the expected number of user edits. The BT1 and BT3 banners beat the BT2 banner in this respect, while the finding on the difference between BT1 and BT3 is inconclusive. BT2 is the only banner from the BT group that performs worse than the GIB_LP and GIB_RG banners in this respect. Also, the GIB_LP and GIB_RG banners do not differ signficantly in the number of expected user edits that they influence.
NOTE. I use the term campaign lift in the following sections to refer to a difference in the probability of user registration in respect to any pair of banners that were used during the ABC2017 campaign. Theoretically, a campaign lift would be the difference (in some KPI) between a group of users who were exposed to the campaign and the control group (no exposure). If we would assess the ABC2017 campaign in this manner then it would be natural to take GIB_RG as a control, and compare all other groups (BT1, BT2, BT3, and GIB_LP) to it; in that case, we would observe that ABC2017 campaign had a lift in respect to the control group in terms of the expected number of user edits.
The following sections provide the results of pairwise Bayesian A/B tests across the campaign banners. TECHNICAL NOTE. Uniform Dirichlet() priors with a concentration parameter of 1 were used to derive the expected user edit distributions, while uniform Beta(1, 1) uninformative priors were used for A/B testing; 1,000,000 Monte Carlo samples from the posteriors were used.
Prepare priors, data, and test functions.
# - Number of Monte Carlo samples:
mcN <- 1e6
# - the dataset
edData <- left_join(editData, userReg,
by = c("rev_user" = "event_userId")) %>%
group_by(event_campaign, edits) %>%
summarise(Count = n())
colnames(edData) <- c('Campaign', 'Edits', 'Count')
# - edData$Match
edData$Match <- edData$Edits
# - max. observed Edits:
maxEdits <- max(edData$Edits)
# - fill in missing edits
campaigns <- unique(edData$Campaign)
nCampaigns <- length(campaigns)
campaigns <- unlist(lapply(campaigns, function(x){
return(rep(x, maxEdits + 1))
}))
edDataCopy <- data.frame(Campaign = campaigns,
Match = rep(seq(0, maxEdits, by = 1), nCampaigns),
stringsAsFactors = F)
edDataCopy <- left_join(edDataCopy, edData,
by = c("Campaign" = "Campaign", "Match" = "Match")
)
edData <- edDataCopy
rm(edDataCopy);
edData$Count[is.na(edData$Count)] <- 0
edData$Edits <- edData$Match
edData$Match <- NULL
# - banner edit probability
bannerEdProb <- edData %>%
group_by(Campaign) %>%
mutate(Prob = Count/sum(Count)) %>%
mutate(Expect = Prob * Edits)
bannerEdProb$Campaign <- toupper(gsub("wmde_abc2017_", "", bannerEdProb$Campaign, fixed = T))
# - true expected edit per banner
campaignTER <- edData %>%
mutate(Ex = Edits * Count) %>%
group_by(Campaign) %>%
summarise(TER = Edits %*% (Count/sum(Count)), SD = sd(Ex))
campaignTER$Campaign <- toupper(gsub("wmde_abc2017_", "", campaignTER$Campaign, fixed = T))
# - N user registrations per Banner
bannerNUser <- regPlotSet %>%
group_by(Campaign) %>%
summarise(Registration = sum(Registrations))
# - posterior expected edit samples:
posteriorEEditSample <- function(alpha, counts, values, samples) {
dirichletSample <- rdirichlet(samples, counts + alpha)
dirichletSample %*% values
}
# - posterior expected edit A/B test:
posterior_EEdit_AB <- function(data, campaignA, campaignB, mcN) {
if (!(campaignA %in% unique(data$Campaign)) |
!(campaignB %in% unique(data$Campaign))) {
stop("Campaign not found.",
call. = TRUE)
} else {
# - prepare res:
res <- list()
# - Uninformative priors
priorA <- rep(1, length(which(data$Campaign %in% campaignA)))
priorB <- rep(1, length(which(data$Campaign %in% campaignB)))
# - Simulate banners:
countsA <- data$Count[which(data$Campaign %in% campaignA)]
countsB <- data$Count[which(data$Campaign %in% campaignB)]
# - edit values:
editValuesA <- data$Edits[which(data$Campaign %in% campaignA)]
editValuesB <- data$Edits[which(data$Campaign %in% campaignB)]
# - posterior expected edits:
posteriorA <- posteriorEEditSample(alpha = priorA,
counts = countsA,
values = editValuesA,
samples = mcN)
posteriorA <- data.frame(posterior = posteriorA)
posteriorB <- posteriorEEditSample(alpha = priorB,
counts = countsB,
values = editValuesB,
samples = mcN)
posteriorB <- data.frame(posterior = posteriorB)
# - res: posteriors
res$posteriorA <- posteriorA
res$posteriorB <- posteriorB
# - res: probability A/B
res$probability <- mean(posteriorA > posteriorB)
# - res: percent difference (campaign Lift)
res$percentDiff <- (posteriorA$posterior - posteriorB$posterior)/posteriorB$posterior*100
area = ifelse(res$percentDiff <= 0, '<= 0', '> 0')
res$percentDiff <- data.frame(percentDiff = res$percentDiff,
area = area,
stringsAsFactors = T)
# - res: true percent difference:
res$t_percentDiff <-
(mean(res$posteriorA$posterior) - mean(res$posteriorB$posterior))/mean(res$posteriorB$posterior)*100
# - out:
return(res)
}
}Expected User Edits per Campaign
The average number of edits per user vs. the campaign via they have registered:
campaignTER$Campaign <- factor(campaignTER$Campaign,
levels = names(campaignChartColors))
colnames(campaignTER) <- c('Campaign', 'Expected', 'S.D.')
ggplot(campaignTER, aes(x = Campaign,
y = Expected,
fill = Campaign,
color = Campaign,
label = round(Expected, 2))) +
geom_bar(stat = "identity",
position = "dodge",
width = .5) +
scale_y_continuous(labels = comma) +
scale_fill_manual("legend", values = campaignChartColors) +
scale_color_manual("legend", values = campaignChartColors) +
geom_label(fill = "white", color = "black") +
ggtitle('Autumn Banner Campaign 2017: User Edits per Campaign') +
theme_minimal() +
theme(axis.text.x = element_text(angle = 0, size = 8, hjust = 1)) +
theme(plot.title = element_text(size = 10)) +
theme(legend.title = element_blank()) +
theme(panel.grid.major.x = element_blank()) +
theme(panel.grid.minor.x = element_blank()) +
theme(panel.background = element_blank())
Expected edits and the respective standard deviations:
campaignTER$Expected <- round(campaignTER$Expected, 2)
campaignTER$`S.D.` <- round(campaignTER$`S.D.`, 2)
knitr::kable(campaignTER, format = "html") %>%
kable_styling(full_width = F, position = "left")| Campaign | Expected | S.D. |
|---|---|---|
| BT1 | 3.03 | 5.01 |
| BT2 | 2.63 | 8.67 |
| BT3 | 4.10 | 8.29 |
| GIB_LP | 1.80 | 4.29 |
| GIB_RG | 2.45 | 4.61 |
Let’s take a closer look upon the distributions of user edits per campaign:
# - the dataset
edDataDist <- left_join(editData, userReg,
by = c("rev_user" = "event_userId")) %>%
dplyr::select(event_campaign, edits)
colnames(edDataDist) <- c('Campaign', 'Edits')
edDataDist$Campaign <- toupper(gsub("wmde_abc2017_", "", edDataDist$Campaign, fixed = T))
edDataDist$Alpha = edDataDist$Edits/max(edDataDist$Edits)
ggplot(edDataDist, aes(x = Campaign, y = Edits,
group = Campaign,
fill = Campaign,
color = Campaign)) +
geom_point(aes(alpha = edDataDist$Alpha),
position = "jitter",
size = 1.5) +
scale_y_continuous(labels = comma) +
scale_fill_manual("legend", values = campaignChartColors) +
scale_color_manual("legend", values = campaignChartColors) +
ggtitle('Autumn Banner Campaign 2017: User Edits Distributions per Campaign') +
theme_minimal() +
theme(axis.text.x = element_text(angle = 0, size = 8, hjust = 1)) +
theme(plot.title = element_text(size = 10)) +
theme(legend.title = element_blank()) +
theme(panel.grid.major.x = element_blank()) +
theme(panel.grid.minor.x = element_blank()) +
theme(panel.background = element_blank()) +
theme(legend.position = 'None')
Note that not too many new users have made any significant number of edits. This fact - the scarcity of avilable data - imposes several constraints upon the present analysis. Please read through carefully and do not jump to conclusions before more data become available.
Evaluation: BT1 vs. BT2
campA <- 'BT1'
campB <- 'BT2'
testAB <- posterior_EEdit_AB(bannerEdProb, campA, campB, mcN = mcN)
# - Probability of campA better than campB:
print(paste('The probability of ',
campA,
' influencing more user edits than ',
campB, ' is : ',
round(testAB$probability, 2),
sep = ""))[1] "The probability of BT1 influencing more user edits than BT2 is : 0.99"# - lift:
print(paste('The percent lift that ',
campA,
' has over ',
campB,
' (',
round(testAB$t_percentDiff, 2),
'%) lies in the interval (',
as.character(round(quantile(testAB$percentDiff$percentDiff, .025)), 2),
'%, ',
as.character(round(quantile(testAB$percentDiff$percentDiff, .975)), 2),
'%) with 95% certainty.',
sep = ""))[1] "The percent lift that BT1 has over BT2 (43.68%) lies in the interval (9%, 90%) with 95% certainty."ggplot(testAB$percentDiff,
aes(x = percentDiff,
group = area,
fill = area)) +
geom_histogram(bins = 1000, alpha = .5) +
scale_y_continuous(labels = comma) +
xlab(paste('(', campA, '-', campB, ')/', campB, sep = "")) + ylab('Density') +
ggtitle(paste(campA, '/', campB, ' Campaign Lift', sep = "")) +
theme_minimal() +
theme(plot.title = element_text(size = 10))
Evaluation: BT1 vs. BT3
campA <- 'BT1'
campB <- 'BT3'
testAB <- posterior_EEdit_AB(bannerEdProb, campA, campB, mcN = mcN)
# - Probability of campA better than campB:
print(paste('The probability of ',
campA,
' influencing more user edits than ',
campB, ' is : ',
round(testAB$probability, 2),
sep = ""))[1] "The probability of BT1 influencing more user edits than BT3 is : 0.43"# - lift:
print(paste('The percent lift that ',
campA,
' has over ',
campB,
' (',
round(testAB$t_percentDiff, 2),
'%) lies in the interval (',
as.character(round(quantile(testAB$percentDiff$percentDiff, .025)), 2),
'%, ',
as.character(round(quantile(testAB$percentDiff$percentDiff, .975)), 2),
'%) with 95% certainty.',
sep = ""))[1] "The percent lift that BT1 has over BT3 (-2.53%) lies in the interval (-25%, 28%) with 95% certainty."ggplot(testAB$percentDiff,
aes(x = percentDiff,
group = area,
fill = area)) +
geom_histogram(bins = 1000, alpha = .5) +
scale_y_continuous(labels = comma) +
xlab(paste('(', campA, '-', campB, ')/', campB, sep = "")) + ylab('Density') +
ggtitle(paste(campA, '/', campB, ' Campaign Lift', sep = "")) +
theme_minimal() +
theme(plot.title = element_text(size = 10))
Evaluation: BT1 vs. GIB_LP
campA <- 'BT1'
campB <- 'GIB_LP'
testAB <- posterior_EEdit_AB(bannerEdProb, campA, campB, mcN = mcN)
# - Probability of campA better than campB:
print(paste('The probability of ',
campA,
' influencing more user edits than ',
campB, ' is : ',
round(testAB$probability, 2),
sep = ""))[1] "The probability of BT1 influencing more user edits than GIB_LP is : 0.79"# - lift:
print(paste('The percent lift that ',
campA,
' has over ',
campB,
' (',
round(testAB$t_percentDiff, 2),
'%) lies in the interval (',
as.character(round(quantile(testAB$percentDiff$percentDiff, .025)), 2),
'%, ',
as.character(round(quantile(testAB$percentDiff$percentDiff, .975)), 2),
'%) with 95% certainty.',
sep = ""))[1] "The percent lift that BT1 has over GIB_LP (12.05%) lies in the interval (-15%, 48%) with 95% certainty."ggplot(testAB$percentDiff,
aes(x = percentDiff,
group = area,
fill = area)) +
geom_histogram(bins = 1000, alpha = .5) +
scale_y_continuous(labels = comma) +
xlab(paste('(', campA, '-', campB, ')/', campB, sep = "")) + ylab('Density') +
ggtitle(paste(campA, '/', campB, ' Campaign Lift', sep = "")) +
theme_minimal() +
theme(plot.title = element_text(size = 10))
Evaluation: BT1 vs. GIB_RG
campA <- 'BT1'
campB <- 'GIB_RG'
testAB <- posterior_EEdit_AB(bannerEdProb, campA, campB, mcN = mcN)
# - Probability of campA better than campB:
print(paste('The probability of ',
campA,
' influencing more user edits than ',
campB, ' is : ',
round(testAB$probability, 2),
sep = ""))[1] "The probability of BT1 influencing more user edits than GIB_RG is : 0.78"# - lift:
print(paste('The percent lift that ',
campA,
' has over ',
campB,
' (',
round(testAB$t_percentDiff, 2),
'%) lies in the interval (',
as.character(round(quantile(testAB$percentDiff$percentDiff, .025)), 2),
'%, ',
as.character(round(quantile(testAB$percentDiff$percentDiff, .975)), 2),
'%) with 95% certainty.',
sep = ""))[1] "The percent lift that BT1 has over GIB_RG (11.47%) lies in the interval (-15%, 47%) with 95% certainty."ggplot(testAB$percentDiff,
aes(x = percentDiff,
group = area,
fill = area)) +
geom_histogram(bins = 1000, alpha = .5) +
scale_y_continuous(labels = comma) +
xlab(paste('(', campA, '-', campB, ')/', campB, sep = "")) + ylab('Density') +
ggtitle(paste(campA, '/', campB, ' Campaign Lift', sep = "")) +
theme_minimal() +
theme(plot.title = element_text(size = 10))
Evaluation: BT2 vs. BT3
campA <- 'BT2'
campB <- 'BT3'
testAB <- posterior_EEdit_AB(bannerEdProb, campA, campB, mcN = mcN)
# - Probability of campA better than campB:
print(paste('The probability of ',
campA,
' influencing more user edits than ',
campB, ' is : ',
round(testAB$probability, 2),
sep = ""))[1] "The probability of BT2 influencing more user edits than BT3 is : 0"# - lift:
print(paste('The percent lift that ',
campA,
' has over ',
campB,
' (',
round(testAB$t_percentDiff, 2),
'%) lies in the interval (',
as.character(round(quantile(testAB$percentDiff$percentDiff, .025)), 2),
'%, ',
as.character(round(quantile(testAB$percentDiff$percentDiff, .975)), 2),
'%) with 95% certainty.',
sep = ""))[1] "The percent lift that BT2 has over BT3 (-32.17%) lies in the interval (-49%, -10%) with 95% certainty."ggplot(testAB$percentDiff,
aes(x = percentDiff,
group = area,
fill = area)) +
geom_histogram(bins = 1000, alpha = .5) +
scale_y_continuous(labels = comma) +
xlab(paste('(', campA, '-', campB, ')/', campB, sep = "")) + ylab('Density') +
ggtitle(paste(campA, '/', campB, ' Campaign Lift', sep = "")) +
theme_minimal() +
theme(plot.title = element_text(size = 10))
Evaluation: BT2 vs. GIB_LP
campA <- 'BT2'
campB <- 'GIB_LP'
testAB <- posterior_EEdit_AB(bannerEdProb, campA, campB, mcN = mcN)
# - Probability of campA better than campB:
print(paste('The probability of ',
campA,
' influencing more user edits than ',
campB, ' is : ',
round(testAB$probability, 2),
sep = ""))[1] "The probability of BT2 influencing more user edits than GIB_LP is : 0.05"# - lift:
print(paste('The percent lift that ',
campA,
' has over ',
campB,
' (',
round(testAB$t_percentDiff, 2),
'%) lies in the interval (',
as.character(round(quantile(testAB$percentDiff$percentDiff, .025)), 2),
'%, ',
as.character(round(quantile(testAB$percentDiff$percentDiff, .975)), 2),
'%) with 95% certainty.',
sep = ""))[1] "The percent lift that BT2 has over GIB_LP (-22.01%) lies in the interval (-42%, 4%) with 95% certainty."ggplot(testAB$percentDiff,
aes(x = percentDiff,
group = area,
fill = area)) +
geom_histogram(bins = 1000, alpha = .5) +
scale_y_continuous(labels = comma) +
xlab(paste('(', campA, '-', campB, ')/', campB, sep = "")) + ylab('Density') +
ggtitle(paste(campA, '/', campB, ' Campaign Lift', sep = "")) +
theme_minimal() +
theme(plot.title = element_text(size = 10))
Evaluation: BT2 vs. GIB_RG
campA <- 'BT2'
campB <- 'GIB_RG'
testAB <- posterior_EEdit_AB(bannerEdProb, campA, campB, mcN = mcN)
# - Probability of campA better than campB:
print(paste('The probability of ',
campA,
' influencing more user edits than ',
campB, ' is : ',
round(testAB$probability, 2),
sep = ""))[1] "The probability of BT2 influencing more user edits than GIB_RG is : 0.04"# - lift:
print(paste('The percent lift that ',
campA,
' has over ',
campB,
' (',
round(testAB$t_percentDiff, 2),
'%) lies in the interval (',
as.character(round(quantile(testAB$percentDiff$percentDiff, .025)), 2),
'%, ',
as.character(round(quantile(testAB$percentDiff$percentDiff, .975)), 2),
'%) with 95% certainty.',
sep = ""))[1] "The percent lift that BT2 has over GIB_RG (-22.43%) lies in the interval (-42%, 3%) with 95% certainty."ggplot(testAB$percentDiff,
aes(x = percentDiff,
group = area,
fill = area)) +
geom_histogram(bins = 1000, alpha = .5) +
scale_y_continuous(labels = comma) +
xlab(paste('(', campA, '-', campB, ')/', campB, sep = "")) + ylab('Density') +
ggtitle(paste(campA, '/', campB, ' Campaign Lift', sep = "")) +
theme_minimal() +
theme(plot.title = element_text(size = 10))
Evaluation: BT3 vs. GIB_LP
campA <- 'BT3'
campB <- 'GIB_LP'
testAB <- posterior_EEdit_AB(bannerEdProb, campA, campB, mcN = mcN)
# - Probability of campA better than campB:
print(paste('The probability of ',
campA,
' influencing more user edits than ',
campB, ' is : ',
round(testAB$probability, 2),
sep = ""))[1] "The probability of BT3 influencing more user edits than GIB_LP is : 0.84"# - lift:
print(paste('The percent lift that ',
campA,
' has over ',
campB,
' (',
round(testAB$t_percentDiff, 2),
'%) lies in the interval (',
as.character(round(quantile(testAB$percentDiff$percentDiff, .025)), 2),
'%, ',
as.character(round(quantile(testAB$percentDiff$percentDiff, .975)), 2),
'%) with 95% certainty.',
sep = ""))[1] "The percent lift that BT3 has over GIB_LP (14.98%) lies in the interval (-13%, 52%) with 95% certainty."ggplot(testAB$percentDiff,
aes(x = percentDiff,
group = area,
fill = area)) +
geom_histogram(bins = 1000, alpha = .5) +
scale_y_continuous(labels = comma) +
xlab(paste('(', campA, '-', campB, ')/', campB, sep = "")) + ylab('Density') +
ggtitle(paste(campA, '/', campB, ' Campaign Lift', sep = "")) +
theme_minimal() +
theme(plot.title = element_text(size = 10))
Evaluation: BT3 vs. GIB_RG
campA <- 'BT3'
campB <- 'GIB_RG'
testAB <- posterior_EEdit_AB(bannerEdProb, campA, campB, mcN = mcN)
# - Probability of campA better than campB:
print(paste('The probability of ',
campA,
' influencing more user edits than ',
campB, ' is : ',
round(testAB$probability, 2),
sep = ""))[1] "The probability of BT3 influencing more user edits than GIB_RG is : 0.83"# - lift:
print(paste('The percent lift that ',
campA,
' has over ',
campB,
' (',
round(testAB$t_percentDiff, 2),
'%) lies in the interval (',
as.character(round(quantile(testAB$percentDiff$percentDiff, .025)), 2),
'%, ',
as.character(round(quantile(testAB$percentDiff$percentDiff, .975)), 2),
'%) with 95% certainty.',
sep = ""))[1] "The percent lift that BT3 has over GIB_RG (14.35%) lies in the interval (-13%, 51%) with 95% certainty."ggplot(testAB$percentDiff,
aes(x = percentDiff,
group = area,
fill = area)) +
geom_histogram(bins = 1000, alpha = .5) +
scale_y_continuous(labels = comma) +
xlab(paste('(', campA, '-', campB, ')/', campB, sep = "")) + ylab('Density') +
ggtitle(paste(campA, '/', campB, ' Campaign Lift', sep = "")) +
theme_minimal() +
theme(plot.title = element_text(size = 10))
Evaluation: GIB_LP vs. GIB_RG
campA <- 'GIB_LP'
campB <- 'GIB_RG'
testAB <- posterior_EEdit_AB(bannerEdProb, campA, campB, mcN = mcN)
# - Probability of campA better than campB:
print(paste('The probability of ',
campA,
' influencing more user edits than ',
campB, ' is : ',
round(testAB$probability, 2),
sep = ""))[1] "The probability of GIB_LP influencing more user edits than GIB_RG is : 0.48"# - lift:
print(paste('The percent lift that ',
campA,
' has over ',
campB,
' (',
round(testAB$t_percentDiff, 2),
'%) lies in the interval (',
as.character(round(quantile(testAB$percentDiff$percentDiff, .025)), 2),
'%, ',
as.character(round(quantile(testAB$percentDiff$percentDiff, .975)), 2),
'%) with 95% certainty.',
sep = ""))[1] "The percent lift that GIB_LP has over GIB_RG (-0.55%) lies in the interval (-25%, 32%) with 95% certainty."ggplot(testAB$percentDiff,
aes(x = percentDiff,
group = area,
fill = area)) +
geom_histogram(bins = 1000, alpha = .5) +
scale_y_continuous(labels = comma) +
xlab(paste('(', campA, '-', campB, ')/', campB, sep = "")) + ylab('Density') +
ggtitle(paste(campA, '/', campB, ' Campaign Lift', sep = "")) +
theme_minimal() +
theme(plot.title = element_text(size = 10))
5.2 Campaign Multi-Channel Attribution Model: Making an Edit
The following model provides for the removal effects of the Campaign channels in respect to whether a user has made any edits at all or not. This procedure instantiates a model of a particular campaign as a directed graph in which every node represents a campaign channel (e.g. a banner, a page view, an act of a user doing something, etc), and then computes the probabilities of transition from one to another channel. In other words, the model estimates the probabilities of taking any of the possible user journeys in the campaign. Once the model is ready, the procedure simulates a large number of user journeys to produce an estimate of the probability of conversion for each of them. In the case of our campaign we consider the event of a user making at least one edit as a conversion. When this step is completed, the procedure starts removing one by one campaign channel from the model, and each time it re-computes the conversion probability to estimate how many conversions would be lost due to the removal of a particular channel. The larger the drop in probability of conversion due to the removal of a particular channel, the larger the removal effect for that channel. Channels with larger removal effects are considered to be more important. The value of the removal effect, being a probability in itself, can vary from 0 to 1.
In this case, the campaign channels are the following events:
BT1- Specific Task Bannerwmde_abc2017_bt1is presented;BT2- Specific Task Bannerwmde_abc2017_bt2is presented;BT3- Specific Task Bannerwmde_abc2017_bt3is presented;GIB- General Inviation Banner -wmde_abc2017_gib_lporwmde_abc2017_gib_rgis presented;TLP- Specific Task PageJetztMitmachenis viewed (note: the same as a banner click on any of the following banners:BT1,BT2,BT3);GLP- General PageMach_mitis viewed; (note: the same as a banner click onGIB_LP);RP- Registration PageBenutzerkonto_anlegenis viewed; (note: encompasses users who transit fromJetztMitmachenorMach_mit, as well as banner clicks onGIB_RG);Reg- The act of user registration;GT- The act of completing the Guided Tour.
Important: unlike in the Bayesian A/B tests that are presented above, where the criterion for pair-wise comparisons among the campaign banners was either the number of users registered (Section 5.1A), or the number of edits made (Section 5.1B), here the criterion (i.e. the definition of conversion, if you prefer) is whether a user has made any edits at all. The reason that motivates this criterion, and not a more strict criterion of making >= 10 edits, is simply because there are only several users who have registered via this campaign and made more than ten edits until now. Removal Effects. The Removal Effect for a campaign channel represents the change in probability that a conversion would obtain if the respective channel was removed from the campaign. Once again, given that conversion here means a user making at least one edit, the removal effects tells us how much would the probability of obtaining at least one edit from a user drop if the respective campaign channel was removed. TECHNICAL NOTE: a Markov model of order 4 was used, with 1e8 total simulation runs from the transition matrix.
Removal Effects
### --- Banner -> Exit paths --- ###
### --- Definition: N(Banner Impressions) - N(BannerClicks == Landing Page Views)
# - define: N(Banner Impressions)
bImp <- banImpSet %>%
group_by(Banner) %>%
summarise(Count = sum(Count))
nBT1 <- bImp$Count[which(bImp$Banner %in% 'BT1')]
nBT2 <- bImp$Count[which(bImp$Banner %in% 'BT2')]
nBT3 <- bImp$Count[which(bImp$Banner %in% 'BT3')]
nGIB <- bImp$Count[which(bImp$Banner %in% 'GIB_LP')] + bImp$Count[which(bImp$Banner %in% 'GIB_RG')]
# - define: N(BannerClicks/PageViews)
bClick <- clickPlotSet %>%
group_by(Source) %>%
summarise(Count = sum(Count))
bClick$Source <- gsub("_click", "", bClick$Source, fixed = T)
# - define: N(BannerImpressions) - N(BannerClicks/PageViews)
nBT1 <- nBT1 - bClick$Count[which(bClick$Source %in% 'BT1')]
nBT2 <- nBT2 - bClick$Count[which(bClick$Source %in% 'BT2')]
nBT3 <- nBT3 - bClick$Count[which(bClick$Source %in% 'BT3')]
nGIB <- nGIB - bClick$Count[which(bClick$Source %in% 'GIB_LP')] + bClick$Count[which(bClick$Source %in% 'GIB_RG')]
### --- Banner -> Landing Page -> Exit paths --- ###
### --- N(Banner Clicks == Landing Page Views) - N(Registration Page Views)
### --- NOTE: TLP == Task Landing Page (JetztMitmachen), GLP == General Landing Page (Mach mit)
nBT1_TLP <- bClick$Count[which(bClick$Source %in% 'BT1')] -
pageSource$n[pageSource$Page %in% 'Spezial:Benutzerkonto_anlegen' & pageSource$Source %in% 'JetztMitmachen_BT1']
nBT2_TLP <- bClick$Count[which(bClick$Source %in% 'BT2')] -
pageSource$n[pageSource$Page %in% 'Spezial:Benutzerkonto_anlegen' & pageSource$Source %in% 'JetztMitmachen_BT2']
nBT3_TLP <- bClick$Count[which(bClick$Source %in% 'BT3')] -
pageSource$n[pageSource$Page %in% 'Spezial:Benutzerkonto_anlegen' & pageSource$Source %in% 'JetztMitmachen_BT3']
nGIB_GLP <- bClick$Count[which(bClick$Source %in% 'GIB_LP')] -
pageSource$n[pageSource$Page %in% 'Spezial:Benutzerkonto_anlegen' & pageSource$Source %in% 'Mach_mit']
### --- Banner (-> Landing Page) -> Registration Page -> Exit paths --- ###
### --- N(Registration Page Views) - N(User Registrations)
bUserReg <- userReg %>%
group_by(event_campaign) %>%
summarise(Count = n())
bUserReg$event_campaign <- toupper(gsub("wmde_abc2017_", "", bUserReg$event_campaign, fixed = T))
colnames(bUserReg)[1] <- 'Banner'
nBT1_TLP_RP <- pageSource$n[pageSource$Page %in% 'Spezial:Benutzerkonto_anlegen' & pageSource$Source %in% 'JetztMitmachen_BT1'] -
bUserReg$Count[bUserReg$Banner %in% 'BT1']
nBT2_TLP_RP <- pageSource$n[pageSource$Page %in% 'Spezial:Benutzerkonto_anlegen' & pageSource$Source %in% 'JetztMitmachen_BT2'] -
bUserReg$Count[bUserReg$Banner %in% 'BT2']
nBT3_TLP_RP <- pageSource$n[pageSource$Page %in% 'Spezial:Benutzerkonto_anlegen' & pageSource$Source %in% 'JetztMitmachen_BT3'] -
bUserReg$Count[bUserReg$Banner %in% 'BT3']
nGIB_GLP_RP <- pageSource$n[pageSource$Page %in% 'Mach_mit' & pageSource$Source %in% 'GIB_LP_click'] -
bUserReg$Count[bUserReg$Banner %in% 'GIB_LP']
nGIB_RP <- pageSource$n[pageSource$Page %in% 'Spezial:Benutzerkonto_anlegen' & pageSource$Source %in% 'GIB_RG_click'] -
bUserReg$Count[bUserReg$Banner %in% 'GIB_RG']
### --- Banner (-> Landing Page) -> Registration Page -> Registration -> Exit --- ###
### --- N(User Registrations) - N(Edited) - N(Completed GT and Not Edited)
userRegGT <- left_join(userReg, gTourData,
by = 'event_userId')
userRegGT <- left_join(userRegGT, editData,
by = c('event_userId' = 'rev_user'))
nBT1_TLP_RP_Reg <- bUserReg$Count[bUserReg$Banner %in% 'BT1'] -
sum((userRegGT$event_campaign %in% 'wmde_abc2017_bt1' & is.na(userRegGT$event_tour)) |
(userRegGT$event_campaign %in% 'wmde_abc2017_bt1' & !is.na(userRegGT$event_tour) & !is.na(userRegGT$edits)))
nBT2_TLP_RP_Reg <- bUserReg$Count[bUserReg$Banner %in% 'BT2'] -
sum((userRegGT$event_campaign %in% 'wmde_abc2017_bt2' & is.na(userRegGT$event_tour)) |
(userRegGT$event_campaign %in% 'wmde_abc2017_bt2' & !is.na(userRegGT$event_tour) & !is.na(userRegGT$edits)))
nBT3_TLP_RP_Reg <- bUserReg$Count[bUserReg$Banner %in% 'BT3'] -
sum((userRegGT$event_campaign %in% 'wmde_abc2017_bt3' & is.na(userRegGT$event_tour)) |
(userRegGT$event_campaign %in% 'wmde_abc2017_bt3' & !is.na(userRegGT$event_tour) & !is.na(userRegGT$edits)))
nGIB_GLP_RP_Reg <- bUserReg$Count[bUserReg$Banner %in% 'GIB_LP'] -
sum((userRegGT$event_campaign %in% 'wmde_abc2017_gib_lp' & is.na(userRegGT$event_tour)) |
(userRegGT$event_campaign %in% 'wmde_abc2017_gib_lp' & !is.na(userRegGT$event_tour) & !is.na(userRegGT$edits)))
nGIB_RP_Reg <- bUserReg$Count[bUserReg$Banner %in% 'GIB_RG'] -
sum((userRegGT$event_campaign %in% 'wmde_abc2017_gib_rg' & is.na(userRegGT$event_tour)) |
(userRegGT$event_campaign %in% 'wmde_abc2017_gib_rg' & !is.na(userRegGT$event_tour) & !is.na(userRegGT$edits)))
### --- Banner (-> Landing Page) -> Registration Page -> Registration -> GT -> Exit --- ###
### --- N(User Registrations) - N(Edited) - N(Completed GT and Not Edited)
nBT1_TLP_RP_Reg_GT <- length(userRegGT$event_userId[(userRegGT$event_campaign %in% 'wmde_abc2017_bt1') &
is.na(userRegGT$event_tour) &
is.na(userRegGT$edits)])
nBT2_TLP_RP_Reg_GT <- length(userRegGT$event_userId[(userRegGT$event_campaign %in% 'wmde_abc2017_bt2') &
is.na(userRegGT$event_tour) &
is.na(userRegGT$edits)])
nBT3_TLP_RP_Reg_GT <- length(userRegGT$event_userId[(userRegGT$event_campaign.x %in% 'wmde_abc2017_bt3') &
is.na(userRegGT$event_tour) &
is.na(userRegGT$edits)])
nGIB_GLP_RP_Reg_GT <- length(userRegGT$event_userId[(userRegGT$event_campaign %in% 'wmde_abc2017_gib_lp') &
is.na(userRegGT$event_tour) &
is.na(userRegGT$edits)])
nGIB_RP_Reg_GT <- length(userRegGT$event_userId[(userRegGT$event_campaign %in% 'wmde_abc2017_gib_rg') &
is.na(userRegGT$event_tour) &
is.na(userRegGT$edits)])
### --- Banner (-> Landing Page) -> Registration Page -> Registration -> GT -> EDIT --- ###
nBT1_TLP_RP_Reg_GT_EDIT <- length(userRegGT$event_userId[(userRegGT$event_campaign %in% 'wmde_abc2017_bt1') &
is.na(userRegGT$event_tour) &
!is.na(userRegGT$edits)])
nBT2_TLP_RP_Reg_GT_EDIT <- length(userRegGT$event_userId[(userRegGT$event_campaign %in% 'wmde_abc2017_bt2') &
is.na(userRegGT$event_tour) &
!is.na(userRegGT$edits)])
nBT3_TLP_RP_Reg_GT_EDIT <- length(userRegGT$event_userId[(userRegGT$event_campaign.x %in% 'wmde_abc2017_bt3') &
is.na(userRegGT$event_tour) &
!is.na(userRegGT$edits)])
nGIB_GLP_RP_Reg_GT_EDIT <- length(userRegGT$event_userId[(userRegGT$event_campaign %in% 'wmde_abc2017_gib_lp') &
is.na(userRegGT$event_tour) &
!is.na(userRegGT$edits)])
nGIB_RP_Reg_GT_EDIT <- length(userRegGT$event_userId[(userRegGT$event_campaign %in% 'wmde_abc2017_gib_rg') &
is.na(userRegGT$event_tour) &
!is.na(userRegGT$edits)])
### --- Banner (-> Landing Page) -> Registration Page -> Registration -> EDIT --- ###
nBT1_TLP_RP_Reg_EDIT <- length(userRegGT$event_userId[(userRegGT$event_campaign %in% 'wmde_abc2017_bt1') &
!is.na(userRegGT$event_tour) &
!is.na(userRegGT$edits)])
nBT2_TLP_RP_Reg_EDIT <- length(userRegGT$event_userId[(userRegGT$event_campaign %in% 'wmde_abc2017_bt2') &
!is.na(userRegGT$event_tour) &
!is.na(userRegGT$edits)])
nBT3_TLP_RP_Reg_EDIT <- length(userRegGT$event_userId[(userRegGT$event_campaign %in% 'wmde_abc2017_bt3') &
!is.na(userRegGT$event_tour) &
!is.na(userRegGT$edits)])
nGIB_GLP_RP_Reg_EDIT <- length(userRegGT$event_userId[(userRegGT$event_campaign %in% 'wmde_abc2017_gib_lp') &
!is.na(userRegGT$event_tour) &
!is.na(userRegGT$edits)])
nGIB_RP_Reg_EDIT <- length(userRegGT$event_userId[(userRegGT$event_campaign %in% 'wmde_abc2017_gib_rg') &
!is.na(userRegGT$event_tour) &
!is.na(userRegGT$edits)])
### --- dataset
mcaData <- data.frame(path = c(deparse(substitute(nBT1)),
deparse(substitute(nBT2)),
deparse(substitute(nBT3)),
deparse(substitute(nGIB)),
deparse(substitute(nBT1_TLP)),
deparse(substitute(nBT2_TLP)),
deparse(substitute(nBT3_TLP)),
deparse(substitute(nGIB_GLP)),
deparse(substitute(nBT1_TLP_RP)),
deparse(substitute(nBT2_TLP_RP)),
deparse(substitute(nBT3_TLP_RP)),
deparse(substitute(nGIB_GLP_RP)),
deparse(substitute(nGIB_RP)),
deparse(substitute(nBT1_TLP_RP_Reg)),
deparse(substitute(nBT2_TLP_RP_Reg)),
deparse(substitute(nBT3_TLP_RP_Reg)),
deparse(substitute(nGIB_GLP_RP_Reg)),
deparse(substitute(nGIB_RP_Reg)),
deparse(substitute(nBT1_TLP_RP_Reg_GT)),
deparse(substitute(nBT2_TLP_RP_Reg_GT)),
deparse(substitute(nBT3_TLP_RP_Reg_GT)),
deparse(substitute(nGIB_GLP_RP_Reg_GT)),
deparse(substitute(nGIB_RP_Reg_GT)),
deparse(substitute(nBT1_TLP_RP_Reg_GT_EDIT)),
deparse(substitute(nBT2_TLP_RP_Reg_GT_EDIT)),
deparse(substitute(nBT3_TLP_RP_Reg_GT_EDIT)),
deparse(substitute(nGIB_GLP_RP_Reg_GT_EDIT)),
deparse(substitute(nGIB_RP_Reg_GT_EDIT)),
deparse(substitute(nBT1_TLP_RP_Reg_EDIT)),
deparse(substitute(nBT2_TLP_RP_Reg_EDIT)),
deparse(substitute(nBT3_TLP_RP_Reg_EDIT)),
deparse(substitute(nGIB_GLP_RP_Reg_EDIT)),
deparse(substitute(nGIB_RP_Reg_EDIT))
),
total_conversions = c(0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
nBT1_TLP_RP_Reg_GT_EDIT, nBT2_TLP_RP_Reg_GT_EDIT, nBT3_TLP_RP_Reg_GT_EDIT,
nGIB_GLP_RP_Reg_GT_EDIT, nGIB_RP_Reg_GT_EDIT, nBT1_TLP_RP_Reg_EDIT,
nBT2_TLP_RP_Reg_EDIT, nBT3_TLP_RP_Reg_EDIT, nGIB_GLP_RP_Reg_EDIT,
nGIB_RP_Reg_EDIT
),
total_null = c(nBT1, nBT2, nBT3, nGIB, nBT1_TLP, nBT2_TLP, nBT3_TLP, nGIB_GLP, nBT1_TLP_RP,
nBT2_TLP_RP, nBT3_TLP_RP, nGIB_GLP_RP, nGIB_RP,
nBT1_TLP_RP_Reg, nBT2_TLP_RP_Reg, nBT3_TLP_RP_Reg, nGIB_GLP_RP_Reg, nGIB_RP_Reg,
nBT1_TLP_RP_Reg_GT, nBT2_TLP_RP_Reg_GT, nBT3_TLP_RP_Reg_GT, nGIB_GLP_RP_Reg_GT,
nGIB_RP_Reg_GT, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0
),
stringsAsFactors = F)
# - correct paths:
mcaData$path <- gsub("n", "", mcaData$path, fixed = T)
mcaData$path <- gsub("_", " > ", mcaData$path, fixed = T)
editEnds <- which(grepl("EDIT", mcaData$path, fixed = T))
for (i in 1:length(editEnds)) {
wPath <- which(mcaData$path %in% gsub(" > EDIT", "", mcaData$path[editEnds[i]], fixed = T))
wOut <- which(mcaData$path == mcaData$path[editEnds[i]])
wPath <- setdiff(wPath, wOut)
mcaData$total_conversions[wPath] <- mcaData$total_conversions[wOut]
}
mcaData <- mcaData[-which(grepl("EDIT", mcaData$path, fixed = T)), ]
### --- MCA model
abc2017Model <- markov_model(mcaData,
var_path = "path",
var_conv = "total_conversions",
var_null = "total_null",
order = 4,
nsim = 1e8,
out_more = T)
# - collect removal effects for the next plot:
re4order <- abc2017Model$removal_effects$removal_effects
### --- Removal Effects:
re <- as.data.frame(abc2017Model$removal_effects)
colnames(re) <- c('Channel', 'Removal Effect')
re$Channel <- factor(re$Channel, levels = as.character(abc2017Model$removal_effects$channel_name))
gplot <- ggplot(data = re,
aes(x = Channel,
y = `Removal Effect`,
label = round(`Removal Effect`, 2))
) +
geom_bar(width = .1, color = "darkblue", fill = "white", stat = "identity") +
geom_label(size = 3) +
scale_y_continuous(labels = comma) +
xlab('Campaign Channel') + ylab('Removal Effect') +
ylim(c(0, 1)) +
ggtitle('Campaign Multi-Channel Attribution: Removal Effects') +
theme_minimal() +
theme(plot.title = element_text(size = 10))Scale for 'y' is already present. Adding another scale for 'y', which will replace the existing scale.suppressWarnings(print(gplot))
Campaign Transition Graph
Each node in the following graph represents a particular campaign channel. The edges of the graph are labeled by the respective transition probabilities between the channels. The size of the node corresponds to its removal effect. TECHNICAL NOTE: the removal effects are derived from a Markov model of order 4, while the transitional probabilities are derived directly from the 1st order model.
### --- MCA model: 1st order for channel-to-channel transitions
abc2017Model <- markov_model(mcaData,
var_path = "path",
var_conv = "total_conversions",
var_null = "total_null",
order = 1,
out_more = T)
### --- plot w. {igraph}
abc2017Net <- data.frame(ougoing = abc2017Model$transition_matrix$channel_from,
incoming = abc2017Model$transition_matrix$channel_to,
stringsAsFactors = F)
abc2017Net$ougoing <- sapply(abc2017Net$ougoing, function(x) {
ch <- gsub("(start)", "START", fixed = T, x)
ch <- gsub("(null)", "EXIT", fixed = T, ch)
ch <- gsub("(conversion)", "EDIT", fixed = T, ch)
ch
})
abc2017Net$incoming <- sapply(abc2017Net$incoming, function(x) {
ch <- gsub("(start)", "START", fixed = T, x)
ch <- gsub("(null)", "EXIT", fixed = T, ch)
ch <- gsub("(conversion)", "EDIT", fixed = T, ch)
ch
})
abc2017Net <- graph.data.frame(abc2017Net,
directed = T)
E(abc2017Net)$label <- round(abc2017Model$transition_matrix$transition_probability, 2)
V(abc2017Net)$color <- c('white',
'indianred1', 'indianred2', 'indianred3', 'cadetblue',
'red', 'blue', 'yellow', 'orange', 'green',
'white', 'white')
V(abc2017Net)$size <- c(20, re4order*40, 20, 20)
V(abc2017Net)$frame.color <- 'white'
# - plot w. {igraph}
coords <- layout_(abc2017Net, as_tree())
par(mai=c(rep(0,4)))
plot(abc2017Net,
layout = coords,
edge.width = .75,
edge.color = "grey",
edge.arrow.size = 0.35,
edge.curved = 0.6,
edge.label.family = "sans",
edge.label.color = "black",
edge.label.cex = .6,
vertex.shape = "circle",
vertex.label.color = "black",
vertex.label.font = 1,
vertex.label.family = "sans",
vertex.label.cex = .75,
vertex.label.dist = .25,
vertex.label.dist = .45,
rescale = F,
xlim = c(-1, 1),
ylim = c(0, 4),
margin = c(rep(0,4)))
5.3 Campaign Multi-Channel Attribution Model: User Registration
TECHNICAL NOTE: a Markov model of order 4 was used, with 1e8 total simulation runs from the transition matrix.
Removal Effects
### --- Banner -> Exit paths --- ###
### --- Definition: N(Banner Impressions) - N(BannerClicks == Landing Page Views)
# - define: N(Banner Impressions)
bImp <- banImpSet %>%
group_by(Banner) %>%
summarise(Count = sum(Count))
nBT1 <- bImp$Count[which(bImp$Banner %in% 'BT1')]
nBT2 <- bImp$Count[which(bImp$Banner %in% 'BT2')]
nBT3 <- bImp$Count[which(bImp$Banner %in% 'BT3')]
nGIB <- bImp$Count[which(bImp$Banner %in% 'GIB_LP')] + bImp$Count[which(bImp$Banner %in% 'GIB_RG')]
# - define: N(BannerClicks/PageViews)
bClick <- clickPlotSet %>%
group_by(Source) %>%
summarise(Count = sum(Count))
bClick$Source <- gsub("_click", "", bClick$Source, fixed = T)
# - define: N(BannerImpressions) - N(BannerClicks/PageViews)
nBT1 <- nBT1 - bClick$Count[which(bClick$Source %in% 'BT1')]
nBT2 <- nBT2 - bClick$Count[which(bClick$Source %in% 'BT2')]
nBT3 <- nBT3 - bClick$Count[which(bClick$Source %in% 'BT3')]
nGIB <- nGIB - bClick$Count[which(bClick$Source %in% 'GIB_LP')] + bClick$Count[which(bClick$Source %in% 'GIB_RG')]
### --- Banner -> Landing Page -> Exit paths --- ###
### --- N(Banner Clicks == Landing Page Views) - N(Registration Page Views)
### --- NOTE: TLP == Task Landing Page (JetztMitmachen), GLP == General Landing Page (Mach mit)
nBT1_TLP <- bClick$Count[which(bClick$Source %in% 'BT1')] -
pageSource$n[pageSource$Page %in% 'Spezial:Benutzerkonto_anlegen' & pageSource$Source %in% 'JetztMitmachen_BT1']
nBT2_TLP <- bClick$Count[which(bClick$Source %in% 'BT2')] -
pageSource$n[pageSource$Page %in% 'Spezial:Benutzerkonto_anlegen' & pageSource$Source %in% 'JetztMitmachen_BT2']
nBT3_TLP <- bClick$Count[which(bClick$Source %in% 'BT3')] -
pageSource$n[pageSource$Page %in% 'Spezial:Benutzerkonto_anlegen' & pageSource$Source %in% 'JetztMitmachen_BT3']
nGIB_GLP <- bClick$Count[which(bClick$Source %in% 'GIB_LP')] -
pageSource$n[pageSource$Page %in% 'Spezial:Benutzerkonto_anlegen' & pageSource$Source %in% 'Mach_mit']
### --- Banner (-> Landing Page) -> Registration Page -> Exit paths --- ###
### --- N(Registration Page Views) - N(User Registrations)
bUserReg <- userReg %>%
group_by(event_campaign) %>%
summarise(Count = n())
bUserReg$event_campaign <- toupper(gsub("wmde_abc2017_", "", bUserReg$event_campaign, fixed = T))
colnames(bUserReg)[1] <- 'Banner'
nBT1_TLP_RP <- pageSource$n[pageSource$Page %in% 'Spezial:Benutzerkonto_anlegen' & pageSource$Source %in% 'JetztMitmachen_BT1'] -
bUserReg$Count[bUserReg$Banner %in% 'BT1']
nBT2_TLP_RP <- pageSource$n[pageSource$Page %in% 'Spezial:Benutzerkonto_anlegen' & pageSource$Source %in% 'JetztMitmachen_BT2'] -
bUserReg$Count[bUserReg$Banner %in% 'BT2']
nBT3_TLP_RP <- pageSource$n[pageSource$Page %in% 'Spezial:Benutzerkonto_anlegen' & pageSource$Source %in% 'JetztMitmachen_BT3'] -
bUserReg$Count[bUserReg$Banner %in% 'BT3']
nGIB_GLP_RP <- pageSource$n[pageSource$Page %in% 'Mach_mit' & pageSource$Source %in% 'GIB_LP_click'] -
bUserReg$Count[bUserReg$Banner %in% 'GIB_LP']
nGIB_RP <- pageSource$n[pageSource$Page %in% 'Spezial:Benutzerkonto_anlegen' & pageSource$Source %in% 'GIB_RG_click'] -
bUserReg$Count[bUserReg$Banner %in% 'GIB_RG']
### --- Banner (-> Landing Page) -> Registration Page -> Registration
nBT1_TLP_RP_Reg <- regData$Registrations[which(regData$Campaign %in% 'BT1')]
nBT2_TLP_RP_Reg <- regData$Registrations[which(regData$Campaign %in% 'BT2')]
nBT3_TLP_RP_Reg <- regData$Registrations[which(regData$Campaign %in% 'BT3')]
nGIB_GLP_RP_Reg <- regData$Registrations[which(regData$Campaign %in% 'GIB_LP')]
nGIB_RP_Reg <- regData$Registrations[which(regData$Campaign %in% 'GIB_RG')]
### --- dataset
mcaData <- data.frame(path = c(deparse(substitute(nBT1)),
deparse(substitute(nBT2)),
deparse(substitute(nBT3)),
deparse(substitute(nGIB)),
deparse(substitute(nBT1_TLP)),
deparse(substitute(nBT2_TLP)),
deparse(substitute(nBT3_TLP)),
deparse(substitute(nGIB_GLP)),
deparse(substitute(nBT1_TLP_RP)),
deparse(substitute(nBT2_TLP_RP)),
deparse(substitute(nBT3_TLP_RP)),
deparse(substitute(nGIB_GLP_RP)),
deparse(substitute(nGIB_RP)),
deparse(substitute(nBT1_TLP_RP_Reg)),
deparse(substitute(nBT2_TLP_RP_Reg)),
deparse(substitute(nBT3_TLP_RP_Reg)),
deparse(substitute(nGIB_GLP_RP_Reg)),
deparse(substitute(nGIB_RP_Reg))
),
total_conversions = c(0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
nBT1_TLP_RP_Reg, nBT2_TLP_RP_Reg, nBT1_TLP_RP_Reg, nGIB_GLP_RP_Reg, nGIB_RP_Reg
),
total_null = c(nBT1, nBT2, nBT3, nGIB, nBT1_TLP, nBT2_TLP, nBT3_TLP, nGIB_GLP, nBT1_TLP_RP,
nBT2_TLP_RP, nBT3_TLP_RP, nGIB_GLP_RP, nGIB_RP,
0, 0, 0, 0, 0
),
stringsAsFactors = F)
# - correct paths:
mcaData$path <- gsub("n", "", mcaData$path, fixed = T)
mcaData$path <- gsub("_", " > ", mcaData$path, fixed = T)
editEnds <- which(grepl("Reg", mcaData$path, fixed = T))
for (i in 1:length(editEnds)) {
wPath <- which(mcaData$path %in% gsub(" > Reg", "", mcaData$path[editEnds[i]], fixed = T))
wOut <- which(mcaData$path == mcaData$path[editEnds[i]])
wPath <- setdiff(wPath, wOut)
mcaData$total_conversions[wPath] <- mcaData$total_conversions[wOut]
}
mcaData <- mcaData[-which(grepl("Reg", mcaData$path, fixed = T)), ]
### --- MCA model
abc2017Model <- markov_model(mcaData,
var_path = "path",
var_conv = "total_conversions",
var_null = "total_null",
order = 4,
nsim = 1e8,
out_more = T)
# - collect removal effects for the next plot:
re4order <- abc2017Model$removal_effects$removal_effects
### --- Removal Effects:
re <- as.data.frame(abc2017Model$removal_effects)
colnames(re) <- c('Channel', 'Removal Effect')
re$Channel <- factor(re$Channel, levels = as.character(abc2017Model$removal_effects$channel_name))
gplot <- ggplot(data = re,
aes(x = Channel,
y = `Removal Effect`,
label = round(`Removal Effect`, 2))
) +
geom_bar(width = .1, color = "darkblue", fill = "white", stat = "identity") +
geom_label(size = 3) +
scale_y_continuous(labels = comma) +
xlab('Campaign Channel') + ylab('Removal Effect') +
ylim(c(0, 1)) +
ggtitle('Campaign Multi-Channel Attribution: Removal Effects') +
theme_minimal() +
theme(plot.title = element_text(size = 10))Scale for 'y' is already present. Adding another scale for 'y', which will replace the existing scale.suppressWarnings(print(gplot))
Campaign Transition Graph
Each node in the following graph represents a particular campaign channel. The edges of the graph are labeled by the respective transition probabilities between the channels. The size of the node corresponds to its removal effect. TECHNICAL NOTE: the removal effects are derived from a Markov model of order 4, while the transitional probabilities are derived directly from the 1st order model.
### --- MCA model: 1st order for channel-to-channel transitions
abc2017Model <- markov_model(mcaData,
var_path = "path",
var_conv = "total_conversions",
var_null = "total_null",
order = 1,
out_more = T)
### --- plot w. {igraph}
abc2017Net <- data.frame(ougoing = abc2017Model$transition_matrix$channel_from,
incoming = abc2017Model$transition_matrix$channel_to,
stringsAsFactors = F)
abc2017Net$ougoing <- sapply(abc2017Net$ougoing, function(x) {
ch <- gsub("(start)", "START", fixed = T, x)
ch <- gsub("(null)", "EXIT", fixed = T, ch)
ch <- gsub("(conversion)", "REGISTRATION", fixed = T, ch)
ch
})
abc2017Net$incoming <- sapply(abc2017Net$incoming, function(x) {
ch <- gsub("(start)", "START", fixed = T, x)
ch <- gsub("(null)", "EXIT", fixed = T, ch)
ch <- gsub("(conversion)", "REGISTRATION", fixed = T, ch)
ch
})
abc2017Net <- graph.data.frame(abc2017Net,
directed = T)
E(abc2017Net)$label <- round(abc2017Model$transition_matrix$transition_probability, 2)
V(abc2017Net)$color <- c('white',
'indianred1', 'indianred2', 'indianred3', 'cadetblue',
'red', 'blue', 'yellow',
'white', 'white')
V(abc2017Net)$size <- c(20, re4order*40, 20, 20)
V(abc2017Net)$frame.color <- 'white'
# - plot w. {igraph}
coords <- layout_(abc2017Net, as_tree())
par(mai=c(rep(0,4)))
plot(abc2017Net,
layout = coords,
edge.width = .75,
edge.color = "grey",
edge.arrow.size = 0.35,
edge.curved = 0.6,
edge.label.family = "sans",
edge.label.color = "black",
edge.label.cex = .6,
vertex.shape = "circle",
vertex.label.color = "black",
vertex.label.font = 1,
vertex.label.family = "sans",
vertex.label.cex = .75,
vertex.label.dist = .25,
vertex.label.dist = .45,
rescale = F,
xlim = c(-1, 1),
ylim = c(0, 4),
margin = c(rep(0,4)))
Summary
The landing page for specific tasks (JetztMitmachen, the TLP channel in the graph) and the GIB campaign are essentially no different in respect to how much they influence user registration. We have learned from the A/B tests that no individual BT (i.e. specific task) banner compares to the performance of GIB_RG which leads directly to the registraion page. However, when considered together, the banners leading to the JetztMitmachen have a performance comparable to GIB_RG. The General Invitation landing page Mach_mit lacks such an effect.
5.3 Campaign Evaluation Summary
ASSUMPTIONS as stated in the Campaign KickOff Presentation:
Assumption 1: More users register when given a clear and low level entry task. RESULTS: When comparing individual banner campaigns, A/B testing shows that more users register via the General Invitation Banner campaign, especially when given no intermediate landing page prior to the registration page. However, the
JetztMitmachencampagin in general has a performance comparable to theGIBcampaign, while theJetztMitmachenpage was certainly more important for user registration than the generalMach_mitpage - as we have learned from the campaign Multi-Channel Attribution model.Assumption 2: A landing page with more information before registration is necessary. RESULTS: A/B testing shows that more users register via
GIB_RGbanner campaign that leads directly to the registration page than via theGIB_LPbanner campaign that has an intermediate landing page. However,Assumption2is supported by a high removal effect of theJetztMitmachenpage.Assumption 3: A general invitation has a lower conversion rate than specific invitations to register. RESULTS: The total number of registered users via the
JetztMitmachencampagin (BT1,BT2, andBT3banner campaigns taken together) is 535, while the total number of users registered via the General Invitation campaign is 519, an almost 50-50 split.
NOTE: All these assumptions are valid if we consider the criterion of making an edit at all instead.
SUGGESTIONS
Suggestion No. 1. Remove the
GIB_RGbanner campaign from future campaigns. It drives almost 90% of the traffic towards the registration page while being the least efficient in terms of influencing new user edits at the same time (NOTE: least efficient in terms of the expected number of user edits, not in terms of making any edits at all). That would probably mean thatdewikiwould acquire less new users during the campaign, but again the goal is probably for it to acquire new editors. Or, even better, take a look at mySuggestion No. 2.Suggestion No. 2. Think about the possibility to integrate the campaign content (e.g. what is on the landing pages now) to the registration page directly. Ratio: the
GIB_RGbanner campaign has no intermediate landing page between banner presentation and registration, leading to the highest number of registered new users; on the other hand, those banner campaigns that instantiate a specific task lead to having more user edits on the average than it (in general; this is not valid forBT2). Maybe integrating the campaign content with the registration page can provide a more powerful combination that would affect positively both registration and future editing.Suggestion No. 3. Remove the
Guided Tourfrom our future campaigns; the analysis of its causal power suggests that it has a negative influence towards making at least one edit on behalf of a newly registered user.
6. Post-Campaign Analytics
This section provides several insights that were sought on the behalf of the campaign management team following the end of the Autumn Banner Campaign 2017.
6. 1 10. October 2017: a change in banners occurs. Did it influence (a) the number of user registrations and (b) the number of user edits?
First, let’s have a look at the total number of user registrations daily:
regPlotSetDaily$Date <- factor(regPlotSetDaily$Date, levels = sort(regPlotSetDaily$Date))
ggplot(regPlotSetDaily, aes(x = Date, y = Registrations)) +
geom_bar(stat = "identity",
position = "dodge",
width = .2) +
scale_y_continuous(labels = comma) +
ggtitle('Autumn Banner Campaign 2017: Total User Registrations Daily') +
theme_minimal() +
theme(axis.text.x = element_text(angle = 90, size = 8, hjust = 1)) +
theme(plot.title = element_text(size = 10)) +
theme(legend.title = element_blank()) +
theme(panel.grid.major.x = element_blank()) +
theme(panel.grid.minor.x = element_blank()) +
theme(panel.background = element_blank())
From the chart only we can see a sharp drop in the number of user registrations following 10/09/2017. The drop is obvious even given that there is a noticeable negative trend since the beginning of the campaign (a rather expected finding). However, let’s perform a simple Chi-Square test to compare the number of user registrations: the first five days of the campaign vs. the last four days (using a 5:4 ratio for the expected distribution split).
populationP <- c(5/9, 4/9)
n <- sum(regPlotSetDaily$Registrations)
expectedCounts <- n*populationP
s <- c(sum(regPlotSetDaily$Registrations[1:5]), sum(regPlotSetDaily$Registrations[6:9]))
print(paste("Expected: ", paste(round(expectedCounts, 2), collapse = ", ")))[1] "Expected: 585.56, 468.44"print(paste("Dataset: ", paste(s, collapse = ", ")))[1] "Dataset: 828, 226"chiSq <- sum(((s - expectedCounts)^2)/expectedCounts)
print(paste("Chi-Square Statistics:", chiSq, sep = " "))[1] "Chi-Square Statistics: 225.859772296015"# - degrees of freedom
df <- 2 - 1 # k == 2 == number of categories
print(paste("D.F.:", df, sep = " "))[1] "D.F.: 1"# - Test significance, alpha == .05
sig <- pchisq(chiSq, df, lower.tail=F) # upper tail
print(paste("Type I Error Prob.:", sig, sep = " "))[1] "Type I Error Prob.: 4.7675165058739e-51"The change in banners that took place on 10/10/2017 has probably influenced the number of user registrations in a negative way.
Let’s perform the same check for the number of user edits.
editsDaily <- editGTData %>%
dplyr::select(edits, `timestamp.x`) %>%
group_by(`timestamp.x`) %>%
summarise(Edits = sum(edits))
colnames(editsDaily) <- c('Date', 'Edits')
editsDaily$Date <-factor(editsDaily$Date, levels = sort(editsDaily$Date))
ggplot(editsDaily, aes(x = Date, y = Edits)) +
geom_bar(stat = "identity",
position = "dodge",
width = .2) +
scale_y_continuous(labels = comma) +
ggtitle('Autumn Banner Campaign 2017: Total User Edits Daily') +
theme_minimal() +
theme(axis.text.x = element_text(angle = 90, size = 8, hjust = 1)) +
theme(plot.title = element_text(size = 10)) +
theme(legend.title = element_blank()) +
theme(panel.grid.major.x = element_blank()) +
theme(panel.grid.minor.x = element_blank()) +
theme(panel.background = element_blank())
The situation here is more complicated given the (a) presence of the general negative trend since the onset of the campaign and (b) a local increase in the number of user edits after 10/10/2017. The appropriate strategy would probably call for a time-series de-trending first, followed by the analysis of the random component only; however, we have only nine points in the data set.
populationP <- c(5/9, 4/9)
n <- sum(editsDaily$Edits)
expectedCounts <- n*populationP
s <- c(sum(editsDaily$Edits[1:5]), sum(editsDaily$Edits[6:9]))
print(paste("Expected: ", paste(round(expectedCounts, 2), collapse = ", ")))[1] "Expected: 334.44, 267.56"print(paste("Dataset: ", paste(s, collapse = ", ")))[1] "Dataset: 473, 129"chiSq <- sum(((s - expectedCounts)^2)/expectedCounts)
print(paste("Chi-Square Statistic:", chiSq, sep = " "))[1] "Chi-Square Statistic: 129.153571428571"# - degrees of freedom
df <- 2 - 1 # k == 2 == number of categories
print(paste("D.F.:", df, sep = " "))[1] "D.F.: 1"# - Test significance, alpha == .05
sig <- pchisq(chiSq, df, lower.tail=F) # upper tail
print(paste("Type I Error Prob.:", sig, sep = " "))[1] "Type I Error Prob.: 6.27690084044491e-30"The chi-square test indicates that much less user edits occurring since 10/10/2017. Given the local increase in the number of edits following 10/10/2017, which is probably unusual given the presence of the general negative trend since the onset of the campaign, we cannot rule out the possibility that the banner change on 10/10/2017 has influenced the number of user edits in a positive way.
6. 3 How many reverted edits there were (a) per campaign, and (b) per user?
NOTE: the following Data Acquisition code chunk is not fully reproducible from this Report. The data are collected by running the script abc2017_PROD_RevertedEdits.R on stat1005.eqiad.wmnet, collecting the data as .tsv files, copying manually, and processing locally. Run from stat1005 stat box by executing Rscript /home/goransm/RScripts/abc2017/abc2017_PROD_RevertedEdits.R.
### --- Script: abc2017_PROD_OverallDailyUpdate.R
### --- the following runs on stat1005.eqiad.wmnet
### --- Rscript /home/goransm/RScripts/abc2017/abc2017_PROD_RevertedEdits.R
### --- The script collects and wrangles a dataset for ABC 2017 post-campaign analytics
### --- WMDE Autumn Banner Campaign 2017.
### --- Goran S. Milovanovic, Data Scientist, WMDE
### --- November 06, 2017.
### -----------------------------------------------------------------------------
### 0. Setup
### -----------------------------------------------------------------------------
rm(list = ls())
library(dplyr)
# - get user registration data: abc2017_userRegistrations.tsv
# - then get user IDs from registered:
setwd('/home/goransm/_miscWMDE/abc2017_DataOUT/abc2017_OfficialDatasets/abc2017_DailyUpdate/')
lF <- list.files()
lF <- lF[grepl('userRegistrations', lF, fixed = T)]
userReg <- read.table(lF,
quote = "",
sep = "\t",
header = T,
check.names = F,
stringsAsFactors = F)
userReg <- userReg %>%
dplyr::select(event_userId, event_isSelfMade) %>%
filter(event_isSelfMade == 1)
# - uids:
uid <- userReg$event_userId
# - sql query
sqlQuery <- paste('SELECT rev_user, rev_id, rev_page, rev_timestamp, rev_sha1, rev_content_model, rev_content_format FROM revision WHERE rev_user IN (',
paste(uid, collapse = ", "),
') AND (rev_timestamp >= 20171004220000) AND (rev_timestamp <= 20171014220000);',
sep = "")
mySqlCommand <- paste('mysql -h analytics-store.eqiad.wmnet dewiki -e ',
paste('"', sqlQuery, '" > ', sep = ""),
'/home/goransm/_miscWMDE/abc2017_DataOUT/abc2017_OfficialDatasets/abc2017_DailyUpdate/abc2017_completeUserRevisions.tsv', sep = "")
system(command = mySqlCommand,
wait = TRUE)Analyse reverted edits locally:
userRevisions <- read.table('./_dailyUpdateDATA/abc2017_completeUserRevisions.tsv',
quote = "",
sep = "\t",
header = T,
check.names = F,
stringsAsFactors = F)
userRevisions <- left_join(userRevisions,
userReg,
by = c("rev_user" = "event_userId"))
userRevisions <- userRevisions %>%
filter(!is.na(event_campaign))
# - keep only those users who made any edits at all:
userRevisions <- userRevisions %>%
filter(rev_user %in% editData$rev_user)
# - Note: UTC times, conversion to CET is not necessary here
userRevisions$rev_timestamp <- as.character(userRevisions$rev_timestamp)
revertsPerUser <- lapply(unique(userRevisions$rev_id), function(x) {
dataset <- dplyr::arrange(userRevisions[userRevisions$rev_user == x, ], rev_timestamp)
return(data.frame(userId = x,
revCount = sum(table(dataset$rev_sha1) - 1),
stringsAsFactors = F))
})
revertsPerUser <- rbindlist(revertsPerUser)
sum(revertsPerUser$revCount)[1] 0In conclusion, no edits were reverted.
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