Customer Churn Prediction Model Comparison – Tableau Dashboard
Losing existing customers (churn) is one of the most concerning issues for businesses, directly impacting revenue and growth.
The client wanted to develop churn prediction models using different machine learning algorithms and compare them graphically.
To address this, we developed a Tableau dashboard that visualizes and compares the performance of multiple machine learning models for churn prediction.
By integrating model outputs from Python into Tableau, the solution provided management with a clear, interactive view of churn probabilities across models.
Project Highlights
- Model Integration: Trained and evaluated four machine learning models — Boosted Trees, Logistic Regression, Neural Networks, and Random Forest — using Python with Scikit-learn, TensorFlow, and XGBoost.
- Performance Comparison Dashboard: Designed a Tableau dashboard to visualize churn probabilities across models with bubble plots, logit comparison charts, and color-coded outcomes.
- Interactive Filters: Implemented dynamic filters for gender, seniority, and billing method to allow segmentation of churn insights by customer attributes.
Business Impact
- Optimized Retention Strategy: Empowered the client to adopt the most effective churn prediction model for their business case.
- Targeted Customer Engagement: Identified at-risk groups, enabling more personalized and effective campaigns to retain those customers.
- Decision Support: Provided an intuitive comparison tool that translated complex model outputs into insights accessible to non-technical decision-makers.
Tools & Technologies
- Python: Scikit-learn, TensorFlow, XGBoost for machine learning model development
- Tableau: Dashboard design and interactive visualization