decision-trees-in-r-and-python

Smarter Marketing with Decision Trees: A Practical Introduction Using R and Python

Overview

This repository contains R and Python implementations of a decision tree model designed to predict whether an online shopper will make a purchase. The analysis is based on the Online Shoppers Purchasing Intention Dataset, which includes user behavior data from an e-commerce website. The goal is to explore how decision trees can help marketers understand customer intent and optimize digital strategies.

Both implementations train a classification tree using key behavioral metrics such as PageValues, BounceRates, and ExitRates to distinguish between purchasers and non-purchasers. While the R version works efficiently with minimal tuning, the Python version requires additional adjustments for feature encoding, pruning, and class balancing.

For more information and a complete code walk-through, please check out the complete blog post at https://blog.marketingdatascience.ai.

Features

• R Implementation: Uses the rpart package to build and visualize a decision tree with simple tuning.
• Python Implementation: Uses scikit-learn to construct a decision tree, with more extensive parameter adjustments for improved balance.
• Data Processing: Includes preprocessing steps like categorical encoding and train-test splitting.
• Evaluation Metrics: Generates a confusion matrix, classification report, and feature importance scores for model assessment.
• Visualization: Decision trees are plotted for both R and Python, showing key decision points.

Source code

• decision-trees-customer-intention.R is the main R script
• decision-trees-customer-intention-ver3.py is the equivalent Python program
• I've also included a few versions of the decision tree graphic as described in the blog post

Dataset

This project is based on the Online Shoppers Purchasing Intention Dataset, which contains 12,330 sessions collected from an e-commerce website. The dataset includes 17 behavioral features such as time spent on different types of pages, bounce rates, and traffic source.

Citation:
Sakar, C. O., & Kastro, Y. (2018). Online Shoppers Purchasing Intention Dataset. UCI Machine Learning Repository.
https://doi.org/10.24432/C5F88Q

License

This project is open-source under the MIT License.