Expresso-Customer-Churn-Prediction

This repository explains how to predict customer churn. An Hackathon Organized by Data Science Nigeria(DSN-AI) to help Expresso predict customer Churn. My 2nd place solution , log_loss of 0.246675 on Zindi where the competition was hosted. I've also added a section in the notebook to get a score of 0.246643, which could be the 'unofficial' 1st place solution .

About Expresso:

Expresso is an African telecommunications company that provides customers with airtime and mobile data bundles. The objective of this challenge is to develop a machine learning model to predict the likelihood of each Expresso customer “churning,” i.e. becoming inactive and not making any transactions for 90 days

My Approach

• Handled Missing Values
• Preprocessed Catgegorical variables
• Clustering
• Feature Creation
• KFold Validation
• Model Blending

Improvements that can be made

• Feature Selection
• Handling missing data more efficiently
• Hyper-parameter tuning

Requirements

• pip install requirements.txt

Leaderboard Scores

• Catboost - 0.2466929
• Xgboost - 0.2469854
• Xgboost and Catboost Blended - 0.246643

If you have any questions, comments or concerns, feel free to reach me on linkedin