VN_ecommerce
NguyenLSCMFTU2
10/03/2020
Introduction
This project is intended to compare the number of searching times based on keywords these are Tiki, Sendo, Shopee,Lazada, then understand more and more e-commerce users activites within a week (7 day)
Set up the packages and standardize datasets to visualize
Load packages
suppressMessages(library(tidyverse))## Warning: package 'tidyverse' was built under R version 3.6.2suppressMessages(library(viridis))## Warning: package 'viridis' was built under R version 3.6.2suppressMessages(library(ggsci))## Warning: package 'ggsci' was built under R version 3.6.3suppressMessages(library(lubridate))
suppressMessages(library(stringr))
suppressMessages(library(stringi))
suppressMessages(library(explore))## Warning: package 'explore' was built under R version 3.6.2suppressMessages(library(DT))## Warning: package 'DT' was built under R version 3.6.2suppressMessages(library(Amelia))## Warning: package 'Amelia' was built under R version 3.6.3suppressMessages(library(gridExtra))## Warning: package 'gridExtra' was built under R version 3.6.2theme_set(theme_minimal())Load datasets
These concepts to manipulate datasets are create a list then use the lapply to replicate to optimize time process
dataset<- list()
dataset<- lapply(Sys.glob("*.csv"),read.csv)
"Built function to standardize"## [1] "Built function to standardize"clean_function <- function(dt) {
dt <- dt %>% mutate(date = as.POSIXct(str_replace_all(date,"T"," "), format = "%Y-%m-%d %H"))
}
dataset<-lapply(dataset, clean_function)
lazada<-as.data.frame(dataset[1])%>% mutate(hour=factor(hour(date)))%>% mutate(wday=factor(wday(date)))
sendo<-as.data.frame(dataset[2])%>% mutate(hour=factor(hour(date)))%>% mutate(wday=factor(wday(date)))
shopee<-as.data.frame(dataset[3])%>% mutate(hour=factor(hour(date)))%>% mutate(wday=factor(wday(date)))
tiki<-as.data.frame(dataset[4])%>% mutate(hour=factor(hour(date)))%>% mutate(wday=factor(wday(date)))Let have a look at sample Lazada dataset right now
library(knitr)
knitr::kable(head(lazada))| date | n | hour | wday |
|---|---|---|---|
| 2020-03-03 09:00:00 | 86 | 9 | 3 |
| 2020-03-03 10:00:00 | 87 | 10 | 3 |
| 2020-03-03 11:00:00 | 92 | 11 | 3 |
| 2020-03-03 12:00:00 | 81 | 12 | 3 |
| 2020-03-03 13:00:00 | 84 | 13 | 3 |
| 2020-03-03 14:00:00 | 86 | 14 | 3 |
From these table, you can easily see the main components are time and n (n is the total of searching times in each record (row))
I also standardize the datasets and extract the weekdays and hour for these below purposes
##Check NA values Once more time, be sure that you are comfortable without NA values
#Test na in datasets
missmap(tiki)
missmap(lazada)
missmap(sendo)
missmap(shopee)These result is quite good
#Visualize the datasets
##Histogram in data set
tiki_plot<-tiki%>% ggplot(aes(x=n,y=..density..))+geom_histogram(bins=35,fill="steelblue",col="white")+geom_density(fill="green",alpha=0.3,col="red")+ggtitle(label="Number of Suffering Tiki Website")
lazada_plot<-lazada%>% ggplot(aes(x=n,y=..density..))+geom_histogram(bins=35,fill="steelblue",col="white")+geom_density(fill="green",alpha=0.3,col="red")+ggtitle(label="Number of Suffering lazada Website")
sendo_plot<-sendo%>% ggplot(aes(x=n,y=..density..))+geom_histogram(bins=35,fill="steelblue",col="white")+geom_density(fill="green",alpha=0.3,col="red")+ggtitle(label="Number of Suffering sendo Website")
shopee_plot<-shopee%>% ggplot(aes(x=n,y=..density..))+geom_histogram(bins=35,fill="steelblue",col="white")+geom_density(fill="green",alpha=0.3,col="red")+ggtitle(label="Number of Suffering shopee Website")
grid.arrange(tiki_plot,sendo_plot,shopee_plot,lazada_plot)
From these plot I can conclude that Shopee has the largest number of searching keyword. But bins are not quite same. Likewise, Lazada frequency clearly separates to 2 parts. Tiki and Sendo are quite same day to day
Hour vs Weekday
Tiki
plot1<- tiki%>% group_by(hour)%>%summarise(total=sum(n))%>%ggplot(aes(x=hour,y=total,fill=hour,alpha=0.5))+geom_col()+labs(title="Searching numbers pear hour of tiki",xlabs="Hour",ylabs="Frequency")+scale_color_viridis(discrete=TRUE,option="A")
plot2<-tiki%>% group_by(wday)%>%summarise(total=sum(n))%>%ggplot(aes(x=wday,y=total,fill=wday))+geom_col()+labs(title="Searching numbers pear wday of tiki",xlabs="Wday",ylabs="Frequency")+scale_color_viridis(discrete=TRUE,option="A")
plot3<- tiki%>%group_by(hour,wday)%>%summarise(total=sum(n))%>%ggplot(aes(x=hour,y=wday))+geom_tile(aes(fill=total))+scale_fill_viridis(option="A")+labs(title="Searching numbers pear hour vs wday of tiki",xlabs="Hour",ylabs="Wday")
plot4<- tiki%>% ggplot(aes(x=date,y=n))+geom_line(aes(group=1),color="blue",lwd=3)+labs(title="Searching numbers pear hour vs wday of tiki",xlabs="Date",ylabs="Frequency")
grid.arrange(plot1,plot2,plot3,plot4,nrow=2,ncol=2)
Sendo
plot1<-sendo%>% group_by(hour)%>%summarise(total=sum(n))%>%ggplot(aes(x=hour,y=total,fill=hour,alpha=0.5))+geom_col()+labs(title="Searching numbers pear hour of sendo",xlabs="Hour",ylabs="Frequency")+scale_color_viridis(discrete=TRUE,option="A")
plot2<-sendo%>% group_by(wday)%>%summarise(total=sum(n))%>%ggplot(aes(x=wday,y=total,fill=wday))+geom_col()+labs(title="Searching numbers pear wday of sendo",xlabs="Wday",ylabs="Frequency")+scale_color_viridis(discrete=TRUE,option="A")
plot3<- sendo%>%group_by(hour,wday)%>%summarise(total=sum(n))%>%ggplot(aes(x=hour,y=wday))+geom_tile(aes(fill=total))+scale_fill_viridis(option="A")+labs(title="Searching numbers pear hour vs wday of sendo",xlabs="Hour",ylabs="Wday")
plot4<- sendo%>% ggplot(aes(x=date,y=n))+geom_line(aes(group=1),color="blue",lwd=3)+labs(title="Searching numbers pear hour vs wday of sendo",xlabs="Date",ylabs="Frequency")
grid.arrange(plot1,plot2,plot3,plot4,nrow=2,ncol=2)
Lazada
plot1<-lazada%>% group_by(hour)%>%summarise(total=sum(n))%>%ggplot(aes(x=hour,y=total,fill=hour,alpha=0.5))+geom_col()+labs(title="Searching numbers pear hour of lazada",xlabs="Hour",ylabs="Frequency")+scale_color_viridis(discrete=TRUE,option="A")
plot2<-lazada%>% group_by(wday)%>%summarise(total=sum(n))%>%ggplot(aes(x=wday,y=total,fill=wday))+geom_col()+labs(title="Searching numbers pear wday of lazada",xlabs="Wday",ylabs="Frequency")+scale_color_viridis(discrete=TRUE,option="A")
plot3<- lazada%>%group_by(hour,wday)%>%summarise(total=sum(n))%>%ggplot(aes(x=hour,y=wday))+geom_tile(aes(fill=total))+scale_fill_viridis(option="A")+labs(title="Searching numbers pear hour vs wday of lazada",xlabs="Hour",ylabs="Wday")
plot4<- lazada%>% ggplot(aes(x=date,y=n))+geom_line(aes(group=1),color="blue",lwd=3)+labs(title="Searching numbers pear hour vs wday of lazada",xlabs="Date",ylabs="Frequency")
grid.arrange(plot1,plot2,plot3,plot4,nrow=2,ncol=2)
Shopee
plot1<-shopee%>% group_by(hour)%>%summarise(total=sum(n))%>%ggplot(aes(x=hour,y=total,fill=hour,alpha=0.5))+geom_col()+labs(title="Searching numbers pear hour of shopee",xlabs="Hour",ylabs="Frequency")+scale_color_viridis(discrete=TRUE,option="A")
plot2<-shopee%>% group_by(wday)%>%summarise(total=sum(n))%>%ggplot(aes(x=wday,y=total,fill=wday))+geom_col()+labs(title="Searching numbers pear wday of shopee",xlabs="Wday",ylabs="Frequency")+scale_color_viridis(discrete=TRUE,option="A")
plot3<- shopee%>%group_by(hour,wday)%>%summarise(total=sum(n))%>%ggplot(aes(x=hour,y=wday))+geom_tile(aes(fill=total))+scale_fill_viridis(option="A")+labs(title="Searching numbers pear hour vs wday of shopee",xlabs="Hour",ylabs="Wday")
plot4<- shopee%>%ggplot(aes(x=date,y=n))+geom_line(aes(group=1),color="blue",lwd=3)+labs(title="Searching numbers pear hour vs wday of shopee",xlabs="Date",ylabs="Frequency")
grid.arrange(plot1,plot2,plot3,plot4,nrow=2,ncol=2)