This repository contains Jyputer Notebook for building various ML models for predicting home defaults
based on various customers' data obtained by the bank.
It's part of Kaggle competition (Home Credit Default Risk competition).
At first EDA was done, based on which I tried to do feature engineering.
For classification logistic regression
and random forest
models were trained and evaluated only on training set (test dataset is not available as that's how Kaggle scores your model).
Those models were built using different set of features.
I recommend viewing and running notebook files with Google Collab, so you won't have to manage any of the python requirements compared with running them locally.