A modular and reusable toolkit for performing feature engineering on structured datasets.
This repository provides essential utilities for preprocessing, transforming, and optimizing features for quant finance and machine learning workflows.
Feature engineering is one of the most critical steps in quant projects.
Here provides a clean pipeline and practical examples for:
- 🧹 Data preprocessing：missing value handling, outlier detection
- 📊 Exploratory Feature Analysis：time series analysis, classical technical indices, correlation, visual comparison
- 🔣 Feature Transformation：temporal dimension, Cross-sectional features, interaction, contextual dimension, demensional reduction
- 📐 Advanced Engineering：improved engineering methods based on the previous results and comparisons
- ✂️ Feature Selection：selection based on correlation changes, SHAP from Catboost models
This repository can serve both as a reference and a reusable feature engineering module.
- Data used: data_cp.csv (too big for uploading)
- Notebooks:
- Technical Indices.ipynb
- TimeSeriesAnalysis.ipynb
- FeatureEngeering_basic.ipynb