Advanced Churn Prediction Machine Learning Pipeline

Develop a comprehensive machine learning pipeline for predicting customer churn in a technology product. Create a workflow that combines feature engineering, multiple model types (logistic regression, random forest, gradient boosting), and cross-validation strategies. Include explicit handling for class imbalance, feature importance analysis, and a method to generate interpretable risk scores. The solution should provide both predictive accuracy and actionable insights for customer success teams.