Churn Prediction in the Waze App: Leveraging Feature Engineering and Machine Learning Models

Overview

• This project predicts user churn in the Waze app using tree-based models, focusing on key factors and retention strategies.

Activity Structure

1. Ethical Considerations

• Address biases and fairness in predicting churn.

2. Feature Engineering

• Develop features to enhance model performance.

3. Modeling

• Build, tune, and compare machine learning models.
• Investigate optimal decision thresholds and feature importance.

Summary

1. Objective

• Predict user churn and identify contributing factors.

2. Insights

• Ensure fairness and address biases.
• Mitigate risks of missed churns and unnecessary retention efforts.

3. Recommendation

• Use the model as a decision-support tool for retention strategies.

Packages Used

• pandas, numpy for data manipulation.
• matplotlib.pyplot, seaborn for visualizations.
• sklearn, xgboost for modeling and evaluation.