Context

StreamBox uses a binary classification model to predict which subscribers are likely to churn in the next 30 days so the retention team can send discount offers. A recently deployed logistic regression model looks strong on overall accuracy, but the retention team says too many churners are still being missed.

Current Performance

The Problem

The VP of Growth is questioning whether the new model is actually better, since higher accuracy has not translated into better retention targeting. You need to evaluate the model properly and explain what the metrics imply for business performance.

Requirements

1. Interpret whether the current model is performing well or poorly for this use case.
2. Explain why accuracy may be misleading given the class balance.
3. Use the precision, recall, F1 score, and confusion matrix to diagnose the model's weaknesses.
4. Compare the current model against the previous model and recommend which one should be used.
5. Propose specific next steps to improve evaluation and model performance.

Constraints

• The retention team can contact at most 8,000 users per week.
• Missing a true churner is more costly than sending an unnecessary offer.
• Model retraining is possible monthly, but feature engineering resources are limited.