Open-source solution study
Credit risk prediction workflow
An end-to-end walkthrough of a Home Credit competition solution, from relational data to model ensembles.
- Problem
- How can customer and credit-history tables be turned into a structured prediction and validation workflow?
- Approach
- Relational aggregation, feature engineering, gradient boosting, out-of-fold predictions, and stacking.
- Output
- A technical guide with workflow diagrams and code examples for each stage.
Read the article and code →Methods and code examples
Feature engineering framework
A practical reference connecting automated features, time-series features, interactions, and feature selection.
- Problem
- How can feature construction and selection be organized for structured data and financial risk tasks?
- Approach
- Automated feature generation, rolling statistics, feature interactions, quality checks, and importance-based selection.
- Output
- A structured collection of methods and implementation examples, including financial ratios and behavioral sequences.
Read the article and code →