Summary

Used data preprocessing and a Random Forest model to predict customer churn, achieving a 0.73 ROC AUC score. Insights can guide retention strategies to improve customer loyalty.

​ Solution

Data preprocessing, including handling missing values and encoding categorical features. A Random Forest model was implemented due to its robustness and effectiveness for classification tasks. Despite various attempts at feature engineering and hyperparameter tuning, the model's performance plateaued at a ROC AUC score of 0.73. This score highlights potential churn risks, providing actionable insights for targeted retention strategies to enhance customer engagement and satisfaction.

Approach

Data exploration to understand the dataset and identify key features. Preprocessed the data by encoding categorical variables. A Random Forest model was selected for its accuracy and interpretability. Despite reaching a ROC AUC score of 0.73, this indicated further opportunities for model refinement and deeper data analysis to enhance predictive accuracy.