Market Basket Analysis

Data and Packages

library(DomoR)
library(tidyverse)
library(arules)
library(arulesViz)
#library(iplots)

load("domoCustomer")
load("domoAccessToken")

DomoR::init(domoCustomer, domoAccessToken)

data <- DomoR::fetch("aac8d0e9-e0c2-4e1e-b33f-92edfdecc3f9") %>%
    select(orders_id, products_id)

Create Transaction Data

transactions <- data %>%
    group_by(orders_id) %>%
    mutate(itemsPurchased = n(),
           moreThanOneItem = ifelse(itemsPurchased > 1, TRUE, FALSE)) %>%
    ungroup() %>%
    filter(moreThanOneItem == TRUE) %>%
    select(orders_id, products_id) %>%
    group_by(orders_id) %>%
    nest() %>%
    ungroup() %>%
    select(-orders_id) %>%
    flatten() %>%
    flatten() %>%
    unname()

Create Association Rules and Plot them

association.rules <- apriori(transactions,
                             parameter = list(supp = 0.00001,
                                              conf = 0.1,
                                              maxlen = 2))

## Apriori
## 
## Parameter specification:
##  confidence minval smax arem  aval originalSupport maxtime support minlen
##         0.1    0.1    1 none FALSE            TRUE       5   1e-05      1
##  maxlen target   ext
##       2  rules FALSE
## 
## Algorithmic control:
##  filter tree heap memopt load sort verbose
##     0.1 TRUE TRUE  FALSE TRUE    2    TRUE
## 
## Absolute minimum support count: 3 
## 
## set item appearances ...[0 item(s)] done [0.00s].
## set transactions ...[4226 item(s), 332315 transaction(s)] done [0.28s].
## sorting and recoding items ... [3099 item(s)] done [0.01s].
## creating transaction tree ... done [0.29s].
## checking subsets of size 1 2 done [0.04s].
## writing ... [3505 rule(s)] done [0.02s].
## creating S4 object  ... done [0.06s].

association.rules <- sort(association.rules, by = 'confidence', decreasing = TRUE)

Plot Rules

plot(association.rules, method="graph", engine = "htmlwidget", control = list(verbose = FALSE, max = 1000))

## Warning: Too many rules supplied. Only plotting the best 1000 rules using
## lift (change control parameter max if needed)

Change Into a Data Frame

rulesDF <- as(association.rules, "data.frame")

associationRules <- rulesDF %>%
  mutate(rulesString = as.character(rules),
         rulesStringEdit = str_replace_all(rulesString, "[:punct:]", ""),
         products_id1 = as.integer(str_extract(rulesStringEdit, "^[:digit:]+")),
         products_id2 = as.integer(str_extract(rulesStringEdit, "[:digit:]+$"))) %>%
  select(-rules, -rulesString, -rulesStringEdit) %>%
  filter(!is.na(products_id1), !is.na(products_id2))

Push to the Cloud

load("domoCustomer")
load("domoAccessToken")

DomoR::init(domoCustomer, domoAccessToken)
#DomoR::create(associationRules, name = "Product Association Rules")
DomoR::replace_ds("ac95cae5-b49c-4b20-97b0-9f435debba43", associationRules)