Folder <- "C:/Laboratory/Transparency/PopTrendsB/";
require(plyr);
require(dplyr);
######################################################################
# Read in the three data files.                                      #
######################################################################
Raw <- read.csv(paste0(Folder,"PopTrendsBData1Fixed.csv"),
                header=TRUE,stringsAsFactors=FALSE);
   Fixed <- data.frame(UserID=Raw$UserID,
                       DateBet=as.Date(Raw$DateBet),
                       StakeF=Raw$StakeF,WinF=Raw$WinF,
                       BetsF=Raw$BetsF,stringsAsFactors=FALSE);
   nFixed <- nrow(Fixed);
   c(min(Fixed$DateBet),max(Fixed$DateBet))
Raw <- read.csv(paste0(Folder,"PopTrendsBData2Live.csv"),
                header=TRUE,stringsAsFactors=FALSE);
   Live <- data.frame(UserID=Raw$UserID,
                       DateBet=as.Date(Raw$DateBet),
                       StakeL=Raw$StakeL,WinL=Raw$WinL,
                       BetsL=Raw$BetsL,stringsAsFactors=FALSE);
   nLive <- nrow(Live);
   c(min(Live$DateBet),max(Live$DateBet))
Raw <- read.csv(paste0(Folder,"PopTrendsBData3Aggs.csv"),
                header=TRUE,stringsAsFactors=FALSE);
   Aggs <- data.frame(UserID=Raw$UserID,CountryID=Raw$CountryID,
                      Gender=Raw$Gender,BirthYear=Raw$BirthYear,
                      DateReg=as.Date(Raw$DateReg),
                      TimeReg=Raw$TimeReg,
                      Date1Dep=as.Date(Raw$Date1Dep),
                      Date1Bet=as.Date(Raw$Date1Bet),
                      Date1Spo=as.Date(Raw$Date1Spo),
                      StakeF=Raw$StakeF,StakeL=Raw$StakeL,
                      StakeA=Raw$StakeA,
                      WinF=Raw$WinF,WinL=Raw$WinL,
                      WinA=Raw$WinA,
                      BetsF=Raw$BetsF,BetsL=Raw$BetsL,
                      BetsA=Raw$BetsA,
                      DaysF=Raw$DaysF,DaysL=Raw$DaysL,
                      DaysA=Raw$DaysA,
                      stringsAsFactors=FALSE);
   nAggs <- nrow(Aggs);
   quantile(probs=c(0,1),x=Aggs$UserID)
   quantile(probs=c(0,1),x=Aggs$BirthYear,na.rm=TRUE)
   c(min(Aggs$DateReg),max(Aggs$DateReg))
############################################################
# The statistics in Section 2.1 check out.                 #
############################################################
# The gender counts and percentages check out.
c(nAggs,sum(Aggs$Gender==1),
  round(100*sum(Aggs$Gender==1)/nAggs,digits=1),
  sum(Aggs$Gender==0),
  round(100*sum(Aggs$Gender==0)/nAggs,digits=1))
# The minimum and maximum ages check out.
# The mean and standard deviation are off because here I'm using
#   birth years and the original analyses used birth dates.
c(2005-max(Aggs$BirthYear,na.rm=TRUE),
  2005-min(Aggs$BirthYear,na.rm=TRUE),
  2005-mean(Aggs$BirthYear,na.rm=TRUE),
  sd(Aggs$BirthYear,na.rm=TRUE))
# The number of countries and the percentage of Germans check out.
c(length(unique(Aggs$CountryID)),
  sum(Aggs$CountryID==276),
  round(100*sum(Aggs$CountryID==276)/nAggs,digits=1))
# The counts and percentages of people who played fixed odds
#   and live action sports betting check out.
c(sum(Aggs$BetsF*Aggs$BetsL>0),
  round(100*sum(Aggs$BetsF*Aggs$BetsL>0)/nAggs,digits=1),
  sum(Aggs$BetsL==0),
  round(100*sum(Aggs$BetsL==0)/nAggs,digits=1),
  sum(Aggs$BetsF==0),
  round(100*sum(Aggs$BetsF==0)/nAggs,digits=1));
############################################################
# Construct daily aggregates from Fixed and Live to show   #
#   that they match the numbers in Aggs.                   #
# * We have to remove data from before subscribers' dates  #
#   of first deposit.                                      #
############################################################
FixedC <- merge(Fixed,Aggs[,c("UserID","Date1Dep")],
                all.x=TRUE,all.y=FALSE,by=c("UserID"));
   FixedC <- subset(FixedC,FixedC$DateBet>=FixedC$Date1Dep);
   AggsF2 <- ddply(FixedC,~UserID,summarise,
                   StakeF=sum(StakeF),WinF=sum(WinF),
                   BetsF=sum(BetsF));
   AggsF1 <- subset(Aggs[,c("UserID","StakesF","WinF","BetsF")],
                    Aggs$BetsF>0);
   quantile(AggsF1$StakeF-AggsF2$StakeF,c(0,1))
   quantile(AggsF1$WinF-AggsF2$WinF,c(0,1))
   quantile(AggsF1$BetsF-AggsF2$BetsF,c(0,1))
LiveC <- merge(Live,Aggs[,c("UserID","Date1Dep")],
               all.x=TRUE,all.y=FALSE,by=c("UserID"));
   LiveC <- subset(LiveC,LiveC$DateBet>=LiveC$Date1Dep);
   AggsL2 <- ddply(LiveC,~UserID,summarise,
                   StakeL=sum(StakeL),WinF=sum(WinL),
                   BetsL=sum(BetsL));
   AggsL1 <- subset(Aggs[,c("UserID","StakesL","WinL","BetsL")],
                    Aggs$BetsL>0);
   quantile(AggsL1$StakeL-AggsL2$StakeL,c(0,1))
   quantile(AggsL1$WinL-AggsL2$WinL,c(0,1))
   quantile(AggsL1$BetsL-AggsL2$BetsL,c(0,1))