|
| 1 | +Sys.setenv(RSTUDIO_PANDOC="/usr/lib/rstudio-server/bin/pandoc") |
| 2 | + |
| 3 | +#Set wd |
| 4 | +getwd() |
| 5 | +setwd("/home/bi_user/Automation/Reports/Cohorts") |
| 6 | + |
| 7 | +rm(list = ls(all=TRUE)) |
| 8 | + |
| 9 | +#Provide new paths to libraries |
| 10 | +.libPaths("/home/bi_user/R/x86_64-pc-linux-gnu-library/3.3/") |
| 11 | +#Load libraries |
| 12 | +library(zoo) |
| 13 | +library(DT) |
| 14 | +library(data.table) |
| 15 | +library(dplyr) |
| 16 | +library(DT) |
| 17 | +library(plyr) |
| 18 | +library(htmlwidgets) |
| 19 | + |
| 20 | +#Load data |
| 21 | +load("DataToCohorts.dat") |
| 22 | + |
| 23 | +#Subset the dataframe |
| 24 | +dataset <- subset(orders_data, select = c(3,5:8,13,16)) |
| 25 | + |
| 26 | +#Convert createdAt to date |
| 27 | +dataset$createdAt <- as.Date(dataset$createdAt, format = "%Y-%m-%d") |
| 28 | + |
| 29 | +#Create a datatable |
| 30 | +dataset = data.table(dataset) |
| 31 | + |
| 32 | +#Create a dt with the min created (aka, first order), aggregation by customer |
| 33 | +newdf_firstorder <- dataset[isvalid == 1 , min(createdAt) , by = .(customer)] |
| 34 | + |
| 35 | +#Merge back to dataset |
| 36 | +dataset <- merge(x = dataset, y= newdf_firstorder, by = "customer", all.x = TRUE) |
| 37 | + |
| 38 | +#Rename the first order col |
| 39 | +colnames(dataset)[8] <- c("firstOrder") |
| 40 | + |
| 41 | +#Get first order |
| 42 | +dataset$sinceFirstOrder <- 12 * as.numeric((as.yearmon(dataset$createdAt)-as.yearmon(dataset$firstOrder))) |
| 43 | +#Change to year-month format |
| 44 | +dataset$initial_cohort <- as.yearmon(dataset$firstOrder) |
| 45 | + |
| 46 | +#Create Customer Status, i.e. convert to 0 if New and 1 to Returning |
| 47 | +dataset$BinCustomerStatus = 0 |
| 48 | +dataset$BinCustomerStatus[which(dataset$createdAt > dataset$firstOrder)] = 1 |
| 49 | + |
| 50 | +#Subset dataset with completed state |
| 51 | +nsubset <- dataset[dataset$state == 'completed'] |
| 52 | + |
| 53 | + |
| 54 | +#Check the count of returning customers, note "by" function: aggregate by BOTH sinceFirstOrder and initiacohort |
| 55 | +cohort = arrange(nsubset[BinCustomerStatus==1, .(sum(isvalid),length(unique(customer))), |
| 56 | + by=.(initial_cohort, sinceFirstOrder)],desc(sinceFirstOrder)) |
| 57 | +colnames(cohort) <- c("initialcohort","sinceFirstOrder","orders","customers") |
| 58 | + |
| 59 | +#Create percentages |
| 60 | +cohort$proportion <- round(cohort$orders /cohort$customers,2) |
| 61 | +#Subset the dataset, deleting two columns |
| 62 | +cohort$orders <- NULL |
| 63 | +cohort$customers <- NULL |
| 64 | + |
| 65 | +#Employ the dcast library |
| 66 | +output <- arrange(dcast(cohort, initialcohort~round(sinceFirstOrder), value.var = "proportion"),initialcohort) |
| 67 | +output = data.table(output) |
| 68 | + |
| 69 | +#Make intitial cohort character for displaying purposes |
| 70 | +output$initialcohort <- as.character(output$initialcohort) |
| 71 | + |
| 72 | + |
| 73 | +#Create output |
| 74 | +cohort_table <- datatable(output,extensions = list("Buttons"= NULL), options = list("paging"= F,"searching"=T, |
| 75 | + "autoWidth"= F, dom = 'Bfrtip', |
| 76 | + buttons = c( 'csv','excel'))) |
| 77 | + |
| 78 | + |
| 79 | +saveWidget(cohort_table, file="/home/bi_user/shinyserver/Cohorts/OrdersPerUser_cohorts_global.html",selfcontained = F) |
| 80 | + |
| 81 | + |
| 82 | +########################################################################################################## |
| 83 | +#### NOW BY CITY |
| 84 | +########################################################################################################## |
| 85 | + |
| 86 | +#Create a df from nsubset with regard to cities |
| 87 | +cohortByCities = arrange(nsubset[BinCustomerStatus==1, .(sum(isvalid),length(unique(customer))), |
| 88 | + by=.(initial_cohort, sinceFirstOrder,location)],desc(initial_cohort)) |
| 89 | + |
| 90 | +colnames(cohortByCities) <- c("initial_cohort","sinceFirstOrder","location","orders","customers") |
| 91 | + |
| 92 | +cohortByCities <- data.table(cohortByCities) |
| 93 | +################################## |
| 94 | +#London |
| 95 | +################################## |
| 96 | +london <- cohortByCities[cohortByCities$location == "gb_london"] |
| 97 | +london <- na.omit(london) |
| 98 | +#Create percentages |
| 99 | +london$proportion <- round(london$orders / london$customers,2) |
| 100 | +#Subset the dataset, deleting two columns |
| 101 | +london$orders <- NULL |
| 102 | +london$customers <- NULL |
| 103 | + |
| 104 | + |
| 105 | +#Employ the dcast library |
| 106 | +lon_output <- arrange(dcast(london, initial_cohort~round(sinceFirstOrder), value.var = "proportion"), initial_cohort) |
| 107 | +lon_output = data.table(lon_output) |
| 108 | + |
| 109 | +lon_output$initial_cohort <- as.character(lon_output$initial_cohort) |
| 110 | +#Create output |
| 111 | +lon_table <- datatable(lon_output,extensions = list("Buttons"= NULL), options = list("paging"= F,"searching"=T, |
| 112 | + "autoWidth"= F, dom = 'Bfrtip', |
| 113 | + buttons = c( 'csv','excel'))) |
| 114 | + |
| 115 | +saveWidget(lon_table, file="/home/bi_user/shinyserver/Cohorts/OrdersPerUser_cohorts_London.html",selfcontained = F) |
| 116 | + |
| 117 | + |
| 118 | +################################## |
| 119 | +#Berlin |
| 120 | +################################## |
| 121 | +berlin <- cohortByCities[cohortByCities$location == "de_berlin"] |
| 122 | +berlin <- na.omit(berlin) |
| 123 | +#Create percentages |
| 124 | +berlin$proportion <- round(berlin$orders / berlin$customers,2) |
| 125 | +#Subset the dataset, deleting two columns |
| 126 | +berlin$orders <- NULL |
| 127 | +berlin$customers <- NULL |
| 128 | + |
| 129 | +#Subset everything after launch |
| 130 | +berlin <- berlin[berlin$initial_cohort >= "Jan 2015"] |
| 131 | + |
| 132 | +#Employ the dcast library |
| 133 | +ber_output <- arrange(dcast(berlin, initial_cohort~round(sinceFirstOrder), value.var = "proportion"), initial_cohort) |
| 134 | +ber_output = data.table(ber_output) |
| 135 | + |
| 136 | + |
| 137 | +ber_output$initial_cohort <- as.character(ber_output$initial_cohort) |
| 138 | +#Create output |
| 139 | +ber_table <- datatable(ber_output,extensions = list("Buttons"= NULL), options = list("paging"= F,"searching"=T, |
| 140 | + "autoWidth"= F, dom = 'Bfrtip', |
| 141 | + buttons = c( 'csv','excel'))) |
| 142 | + |
| 143 | +saveWidget(ber_table, file="/home/bi_user/shinyserver/Cohorts/OrdersPerUser_cohorts_Berlin.html",selfcontained = F) |
| 144 | + |
| 145 | + |
| 146 | +################################## |
| 147 | +#Paris |
| 148 | +################################## |
| 149 | +paris <- cohortByCities[cohortByCities$location == "fr_paris"] |
| 150 | +paris <- na.omit(paris) |
| 151 | +#Create percentages |
| 152 | +paris$proportion <- round(paris$orders / paris$customers,2) |
| 153 | +#Subset the dataset, deleting two columns |
| 154 | +paris$orders <- NULL |
| 155 | +paris$customers <- NULL |
| 156 | + |
| 157 | +#Subset everything after launch |
| 158 | +paris <- paris[paris$initial_cohort >= "May 2016"] |
| 159 | + |
| 160 | +#Employ the dcast library |
| 161 | +par_output <- arrange(dcast(paris, initial_cohort~round(sinceFirstOrder), value.var = "proportion"), initial_cohort) |
| 162 | +par_output = data.table(par_output) |
| 163 | + |
| 164 | + |
| 165 | +par_output$initial_cohort <- as.character(par_output$initial_cohort) |
| 166 | +#Create output |
| 167 | +par_table <- datatable(par_output,extensions = list("Buttons"= NULL), options = list("paging"= F,"searching"=T, |
| 168 | + "autoWidth"= F, dom = 'Bfrtip', |
| 169 | + buttons = c( 'csv','excel'))) |
| 170 | + |
| 171 | +saveWidget(par_table, file="/home/bi_user/shinyserver/Cohorts/OrdersPerUser_cohorts_Paris.html",selfcontained = F) |
| 172 | + |
0 commit comments