inst/case_studies/Market Basket Analysis in R.md

In datacamp/rexplorables: Support Authoring of Explorable and NonCoding Exercises

Market Basket Analysis in R: Case Study in using explorable exercise with MCQ

Author: Adel Nehme

Context:

In the course Market Basket Analysis in R, students are tasked in analyzing transactional datasets and compute associations between different items. The most commonly used package to perform such analysis is in the arules package, which takes in datasets of the class transactions and rules and computes interesting relationships between transactions and items in a basket. A companion package used in the course, named arulesViz, lets students visualize these transactions and rules. One function used in this package is the ruleExplorer() function which takes in a rules object and outputs a shiny app with a variety of analytics.

It was a great candidate to be displayed as an explorable with MCQ exercise so that students are prompted to interact with the dashboard while responding to questions as such:

https://www.loom.com/share/34ac9377d26045e79a10afa29215c812

How to:

Even though this process is given in the GitHub documentation of the new exercise types, I will be outlining the steps I took in order to get this particular app running.

1) Make sure the course has a shiny base-image in the course.yml file. 2) Make sure the rexplorables package is installed in the requirements.R file. 3) Make sure XML is installed in the requirements.sh file. 4) Launch local RStudio environment to test out the shiny app. 5) Create a folder to house your app - name this folder chX_exY for better tracking in case you want to have more than one Explorable exercise in the course. 6) Add an app.R file in this folder. 7) Load the code needed to launch the app from A to Z (preprocessing data, to inserting the data in a shiny app) in the app.R file while making sure the rexplorables package is loaded in the file. Note that the ruleExplorer() function takes in a rules object and outputs a shiny app. The explorable exercise types do not require any changes to the source code to adapt any of the original source code of the function, and the code I used was just the following: ```R # Loading required packages - including rexplorables library(rexplorables) library(arules) library(tidyverse) library(arulesViz)

Transformations needed to make sure association rules are created

Online_Retail_2011_Q1 = read.csv("https://assets.datacamp.com/production/repositories/5023/datasets/8759a37fd5570a976159ff8fcf97af521bd3c778/Online_Retail_2011_Q1_sub.csv", header = TRUE)

Online_Retail_clean = Online_Retail_2011_Q1 %>% filter(complete.cases(.))

data_list = split(Online_Retail_clean$Description, Online_Retail_clean$InvoiceNo) Online_trx = as(data_list, "transactions")

rules_online = apriori(Online_trx, parameter = list(supp = 0.01, conf = 0.8, minlen = 2))

Actual code that launches shiny app

ruleExplorer(rules_online)

8) After having made sure the app works fine in your local set up, save it in the `chX_exY` folder and zip the folder - for simplicity, name the zip file `chX_exY` as well.
9) Upload the zip file to teach, and implement the following code in the PEC of the explorable exercise of choice: <br>
```R
# Link to the uploaded zip file in assets
url <- "https://assets.datacamp.com/production/repositories/5023/datasets/04f991c1cf726753a3ba0b98c4c2d10a16bb7850/ch1_ex2.zip"

# Copy the URL as an explorable exercise
rexplorables::copy_explorables(url)

# Display app.R file inside of the explorable folder
displayPage(“ch1_ex2/')

Note that it's preferred to copy-paste the following lines of code in the requirements.R file before soft launch to make sure the app is saved in the docker image and the web app won't have to be downloaded each time

# Link to the uploaded zip file in assets
url <- "https://assets.datacamp.com/production/repositories/5023/datasets/04f991c1cf726753a3ba0b98c4c2d10a16bb7850/ch1_ex2.zip"

# Copy the URL as an explorable exercise
rexplorables::copy_explorables(url)

If you have any questions on how to get this done, feel free to slack me or email me!

datacamp/rexplorables documentation built on June 6, 2021, 4:12 p.m.