Performative Prediction project - Real Data

This repo is part of 236608 project at the Technion. The dataset was taken from GiveMeSomeCredit at Kaggle: https://www.kaggle.com/c/GiveMeSomeCredit

Installation

The following packages are required:

torch
numpy
matplotlib # for visualization
sklearn
pandas

Usage

For transition 1 (the one used in performative prediction paper):

./run_script_tran1.sh i # i is the optimizer index - 1~7

For transition 2 (uses all of the attributes for the distribution shift):

./run_script_tran2.sh i # i is the optimizer index - 1~7

For both transitions:

./run_script.sh i # i is the optimizer index - 1~7

The indexes refer to the following:

1 -> RRM
2 -> RRM + ADAM (0.1)
3 -> RRM + ADAM (0.1) + lookahead
4 -> RRM + ADAM (0.1) + learned lookahead
5 -> RRM + ADAM (0.9)
6 -> RRM + ADAM (0.9) + lookahead
7 -> RRM + ADAM (0.9) + learned lookahead

License

MIT