How astrozoid2604/maybank_assessment works
This is a banking analytics project to identify existing NORMAL customers who may qualify as hidden AFFLUENT customers and recommend segment-upgrade strategies. The documented workflow covers customer-data exploration and feature selection, data processing (including imputation, encoding, dimensionality reduction, and imbalance handling), predictive modeling and evaluation, then upgrade recommendations. The tree also shows prepared-data, model-output, and recommendation artifacts. No source excerpts were provided, so links between those artifacts are shown as unverified where the README does not establish them.
Main entry points: JAMESLIM_notebook.ipynb EDA and features · DATA_INIT.pkl Initial datasets · dataset Balanced datasets · consolidated_result.pkl Tuning results
Components 10
Customer analysis
- Customer data
- EDA and features
JAMESLIM_notebook.ipynb
Data processing
- Initial datasets
dataset/DATA_INIT.pkl - Balanced datasets
dataset
Predictive modeling
- Model training
JAMESLIM_notebook.ipynb - Tuning results
model_training_tuning_output/consolidated_result.pkl - Result consolidation
Loading_consolidated_result.ipynb
Upgrade strategy
- Upgrade recommendations
recommendation_output - Recommended customers
Finalized_recommended_customers_for_upgrade.xlsx
Other components
- Bank team
Connections 9
- Bank team → Customer data: analyzes
README.md - Customer data → EDA and features: explores
README.md - EDA and features → Initial datasets: prepares
- Initial datasets → Balanced datasets: transforms
- Balanced datasets → Model training: trains on
- Model training → Tuning results: evaluates
- Tuning results → Result consolidation: loads
- Result consolidation → Upgrade recommendations: informs
- Upgrade recommendations → Recommended customers: selects