Skip to content

hannesoehler/kaggle_predicting_students_success

Repository files navigation

Kaggle competition: two models to predict students' success

A transformer model and a LightGBM model to predict how students will perform on future questions. The transformer model reached a AUC ROC score of 0.784 on Kaggle's private leaderboard, the Light GBM a score of 0.801 (top 2%).

Requirements

  • Python 3 / jupyter
  • Accept competition rules on Kaggle's competition webpage.
  • Download the data from the competition wepage to current directory.
  • GPU for the transformer model (recommended)

Installation

pip install -r requirements.txt

About

Kaggle competition: two models to predict students' success

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published