This repository contains my submission to the G Research Financial Forecasting Competition for which I was ranked 33 on the public leaderboard.
The notebooks document:
- Exploratory data analysis
- Feature preprocessing using methods such as tree-based discretisation, target encoding, and entity embeddings
- Feature selection
- Model training (XGboost, LightGBM, Neural Networks)
- Hyperparameter tuning
- Training of entity embeddings of the Stock feature using Keras
The library contains some helper functions for:
- Feature preprocessing
- Model training and validation
- Entity embeddings