Customer Lifecycle Management: Player Churn Analysis & Retention Strategy
Overview
Analyzed player behavior across Summer and Fall seasons to uncover churn patterns. Using data visualization, segmentation, and predictive modeling, I identified key churn drivers and proposed actionable retention strategies.
Exploratory Data Analysis (EDA): Analyzed age, income, player types, session counts, and seasonal differences.
Churn Rate Analysis: Calculated churn rates by customer type, income level, and age group.
Predictive Modeling: Built an XGBoost classifier achieving an AUC of 0.71.
Feature Importance: Identified fallbonus, social activity, and customer type as top predictors of churn.
Retention Strategy Recommendations: Developed targeted strategies based on player segments and churn patterns.
Key Features
Visual Highlights
Bonus programs greatly reduce churn, especially among social players.
Social players have the highest churn rate without incentives.
Newer cohorts (2022+) show higher churn rates.
Targeted interventions by age group and player type proved highly effective.
Key Takeaways
Tools & Technologies
Python: Pandas, Matplotlib, Seaborn, Scikit-learn, XGBoost
Machine Learning: Classification, Evaluation Metrics
Data Visualization
Future Work
Data Enrichment:
Incorporate additional player attributes such as in-game purchases, login frequency, and social connections to improve model accuracy.
Model Optimization:
Experiment with advanced hyperparameter tuning techniques such as Random Search and Bayesian Optimization to further enhance the predictive performance.
Longitudinal Analysis:
Extend the analysis to longer time periods to detect seasonal trends and predict long-term player retention more accurately.
Retention Campaign Simulation:
Simulate different incentive strategies and analyze their projected impact on churn rates before real-world implementation.
Deployment:
Deploy the churn prediction model into a real-time environment where predictions can be used to trigger immediate retention actions.