customer_retention

Machine learning and logistic regression predictions of retained customers

This code runs a customer retention analysis of customers over the course of 6 months. The data include 50,000 customer records in 3 cities, and 12 factors that can be used to predict retention.

The goals of the analysis are:

• Perform an exploratory analysi and visualizations to understand the data.
• Build a model to predict whether or not a customer will be active in their 6th month after joining.

The analysis consists of:

• Exploratory summary stats, pairs plot, select boxplots
• Partition data into training & test sets

Logistic Regression model

• identify key factors
• assess model prediction accuracy
• ROC & test set confusion matrix

Machine Learning: Random Forest model

• assess model prediction accuracy
• Test set confusion matrix
• Variable importance (compare with logistic regression)

The analysis was originally done late 2016. I made minor updates to post it here.