Global Banking Churn Analysis

My goal is to understand and reduce customer churn for our banking products across several foreign markets. I have already compiled a kagal dataset that captures transactional behaviour, limited feedback, and basic demographic information from customers outside our home country. While the primary focus is international, I am open to insights that compare or contrast regions such as Europe, Asia, or North America if that helps surface actionable patterns.

The work involves:
• Exploring the dataset, cleaning it, and engineering features that reflect regional regulations, cultural payment preferences, and usage nuances.
• Building a churn-propensity model that clearly identifies the drivers most relevant to banking customers abroad. Python, R, or another proven machine-learning framework is acceptable as long as the code is reproducible.
• Translating model output into plain-language recommendations I can hand straight to our retention and product teams—specific changes to onboarding, messaging, or fee structures that are likely to keep at-risk customers engaged.
• Packaging everything in a concise report and a notebook or script that can be rerun whenever fresh data arrives.

Acceptance criteria:
1. A validated churn model with performance metrics (ROC-AUC and precision-recall on a held-out set).
2. Ranked list of the top churn drivers with commentary tailored to foreign markets.
3. Practical, region-aware retention actions tied directly to those drivers.
4. Clean, well-commented code and a brief hand-off session to walk through results.

Once delivered, I’ll test the analysis on a new monthly data slice; if the predictive lift holds, we can extend the engagement to ongoing monitoring and A/B experimentation.