/credit-scoring

Machine Learning for credit scoring.

Primary LanguageJupyter Notebook

Machine Learning for credit scoring

Machine Learning and Deep Learning models for credit default prediction.

Setup

The provided Docker file can be used to setup a container with everything that is needed to reproduce the presented experiments:

docker build -t credit-scoring:latest -f Dockerfile .

Make sure to update the -v /home/rr/DevOps/:/home/rr/DevOps parameter and run the container for the first time using:

docker run --gpus 'all,"capabilities=graphics,utility,display,video,compute"' --net host --privileged --name credit-scoring -itu rr -e NVIDIA_VISIBLE_DEVICES=all -e DISPLAY=$DISPLAY -v /tmp/.X11-unix:/tmp/.X11-unix:rw -v /home/rr/DevOps:/home/rr/DevOps credit-scoring /bin/bash

If you get errors like Error: cannot open display, try fixing it by running

xhost local:root

In order to run the Jupyter notebooks, src must be in your Python path:

export PYTHONPATH="${PYTHONPATH}:/home/rr/DevOps/credit-scoring"

Create models directory:

mkdir models

Download datasets:



Start the Jupyter Lab:


cd /home/rr/DevOps/credit-scoring
~/.local/bin/jupyter lab --no-browser --ip "*"