This repository contains the churn prediction analysis for a telecom company. The goal is to predict which customers will churn in the near future using Decision Tree and Naïve Bayes predictive models.
The dataset includes various customer attributes such as daily usage, service calls, and plan subscriptions, which are used to predict churn likelihood.
Two types of predictive models were used:
- Decision Tree
- Best accuracy: 92.32%
- Implemented in Python
- Naïve Bayes
- Best accuracy: 84.79%
- Implemented in SAS
- The most important predictors of customer churn are:
- Day Mins: The number of minutes the customer used the service during daytime.
- CustServ Calls: The number of calls to customer support.
- Int'l Plan: Whether the customer has an international calling plan.
- Focus on improving customer experience for heavy users with high day minutes and those with international plans.
- Implement a flag system in the CRM to monitor customers with increased churn risk.
Ensure you have Python and SAS installed on your machine. Each notebook contains detailed steps to run the models and visualize the results.