WiDS2021

Kaggle WiDS 2021 competition

Some code used for the submissions I've made to the WiDS 2021 competition on Kaggle.

Python libraries used

pandas, numpy, catboost, lightgbm, hyperopt

Motivation for the project

First Kaggle competition, with a friend (InesT).

Files descriptions

• kaggle_setup.py : how to download data from kaggle via Kaggle API for python
• setup.py : needed packages and functions
• WiDS_v1_catboost_RS.ipynb : first submit, catBoost model with random search for parameter tunning
• WiDS_lightGBM_hyperopt_numerical_feat.ipynb : lightGBM model using only numerical features, parameters tunning with hyperopt

Conclusions of the project

Scores obtained on the unlabelled data:

• submission_catboost_RS_180221.csv : 0.84240