R Notebook

This data is downloaded from Kaggle[https://www.kaggle.com/carrie1/ecommerce-data] and its original source is UCI Machine Learning Repository

Loading libraries

library(data.table)

data.table 1.12.6 using 2 threads (see ?getDTthreads).  Latest news: r-datatable.com

Attaching package: 㤼㸱data.table㤼㸲

The following objects are masked from 㤼㸱package:dplyr㤼㸲:

    between, first, last

The following object is masked from 㤼㸱package:purrr㤼㸲:

    transpose

library(dplyr)
library(ggplot2)
library(stringr)
library(tidyr)
library(tidyverse)
library(arules)

Attaching package: 㤼㸱arules㤼㸲

The following object is masked from 㤼㸱package:car㤼㸲:

    recode

The following object is masked from 㤼㸱package:dplyr㤼㸲:

    recode

The following objects are masked from 㤼㸱package:base㤼㸲:

    abbreviate, write

library(arulesViz)

Loading required package: grid
Registered S3 method overwritten by 'seriation':
  method         from 
  reorder.hclust gclus
Registered S3 methods overwritten by 'htmltools':
  method               from         
  print.html           tools:rstudio
  print.shiny.tag      tools:rstudio
  print.shiny.tag.list tools:rstudio
Registered S3 method overwritten by 'htmlwidgets':
  method           from         
  print.htmlwidget tools:rstudio

library(knitr)
library(gridExtra)

Attaching package: 㤼㸱gridExtra㤼㸲

The following object is masked from 㤼㸱package:randomForest㤼㸲:

    combine

The following object is masked from 㤼㸱package:dplyr㤼㸲:

    combine

library(lubridate)

Attaching package: 㤼㸱lubridate㤼㸲

The following objects are masked from 㤼㸱package:data.table㤼㸲:

    hour, isoweek, mday, minute, month, quarter, second, wday, week, yday, year

The following object is masked from 㤼㸱package:base㤼㸲:

    date

library(plyr)

-------------------------------------------------------------------------------------------------------------------------------------------------------------------------
You have loaded plyr after dplyr - this is likely to cause problems.
If you need functions from both plyr and dplyr, please load plyr first, then dplyr:
library(plyr); library(dplyr)
-------------------------------------------------------------------------------------------------------------------------------------------------------------------------

Attaching package: 㤼㸱plyr㤼㸲

The following object is masked from 㤼㸱package:lubridate㤼㸲:

    here

The following objects are masked from 㤼㸱package:dplyr㤼㸲:

    arrange, count, desc, failwith, id, mutate, rename, summarise, summarize

The following object is masked from 㤼㸱package:purrr㤼㸲:

    compact

Loading data

dim(df_data)

[1] 406829      8

Data Cleaning and Feature Engineering

#Removing NAs
df_data <- df_data %>%
  drop_na()


#Converting everything to factors
df_data <- df_data %>% 
  mutate(InvoiceNo=as.factor(InvoiceNo), StockCode=as.factor(StockCode), 
         InvoiceDate=as.Date(InvoiceDate, '%m/%d/%Y %H:%M'), CustomerID=as.factor(CustomerID), 
         Country=as.factor(Country))

Creating Recency,Frequency and Monetory variables for every user for RFM Analysis !

df_RFM <- df_data %>% 
  group_by(CustomerID) %>% 
  summarise(recency=as.numeric(as.Date("2012-01-01")-max(InvoiceDate)),
            frequenci=n_distinct(InvoiceNo), monitery= sum(total_dolar)/n_distinct(InvoiceNo)) 

#lets see how these variables are distributed

ggplot(df_RFM,aes(x=recency)) + geom_histogram()

ggplot(df_RFM,aes(x=frequenci)) + geom_histogram() 

ggplot(df_RFM,aes(x=monitery)) + geom_histogram()

NA
NA

Now these users can be clustered into different group by :
1.Scoring them on quantile values of Recency,Frequency and Monitory 2.K-means clustering 3.Heirarchical clustering

Lets choose heirarchial clustering here!

Market Basket Analysis to see which products were sold together! This can be combined with the targetting/discounting strategy above to show the products which have high frequency of being bought together!

class(tr)

[1] "transactions"
attr(,"package")
[1] "arules"

Lets see how does our products look

itemFrequencyPlot(tr,topN=20,type="absolute",col=brewer.pal(8,'Pastel2'), main="Absolute Item Frequency Plot")

#Hanging T-Light holder is sold the most followed by Regency Cakestand 3 tier

#Using apriori algorithm to set up association rules 
association.rules <- apriori(tr, parameter = list(supp=0.001, conf=0.8,maxlen=10))

Apriori

Parameter specification:

Algorithmic control:

Absolute minimum support count: 22 

set item appearances ...[0 item(s)] done [0.03s].
set transactions ...[8181 item(s), 22191 transaction(s)] done [0.56s].
sorting and recoding items ... [2623 item(s)] done [0.03s].
creating transaction tree ... done [0.02s].
checking subsets of size 1 2 3 4 5 6 7 8 9 10

Mining stopped (maxlen reached). Only patterns up to a length of 10 returned!

 done [0.64s].
writing ... [49122 rule(s)] done [0.10s].
creating S4 object  ... done [0.08s].

summary(association.rules)

set of 49122 rules

rule length distribution (lhs + rhs):sizes
    2     3     4     5     6     7     8     9    10 
  105  2111  6854 16424 14855  6102  1937   613   121 

   Min. 1st Qu.  Median    Mean 3rd Qu.    Max. 
  2.000   5.000   5.000   5.499   6.000  10.000 

summary of quality measures:
    support           confidence          lift             count       
 Min.   :0.001036   Min.   :0.8000   Min.   :  9.846   Min.   : 23.00  
 1st Qu.:0.001082   1st Qu.:0.8333   1st Qu.: 22.237   1st Qu.: 24.00  
 Median :0.001262   Median :0.8788   Median : 28.760   Median : 28.00  
 Mean   :0.001417   Mean   :0.8849   Mean   : 64.589   Mean   : 31.45  
 3rd Qu.:0.001532   3rd Qu.:0.9259   3rd Qu.: 69.200   3rd Qu.: 34.00  
 Max.   :0.015997   Max.   :1.0000   Max.   :715.839   Max.   :355.00  

mining info:

We can see that Billboard font design & Wrap, Art lights & funk monkey, Swiss roll towel and chocolate spots have very high probability of being sold together, as compared to being sold individually!