MISFEAT Experimentation Framework

This repository provides a framework to run experiments with different models (MISFEAT, MINE, MLP) on various datasets (loan, mobile, attrition, sd1, sd2). It also includes a script to compute metrics for the results.

---

Table of Contents

1. Setup
2. Running Experiments
3. Computing Metrics
4. Example Commands

---

Setup

Prerequisites

• Python 3.8 or higher
• Required Python libraries (install via requirements.txt)

Installation

1. Clone the repository:

   git clone https://github.com/your-repo/misfeat.git
   cd misfeat
   

2. Install dependencies: pip install -r requirements.txt

Running Experiments

Script: run_experiments.sh

First, ensure run_experiments.sh is executable, run this command line: chmod +x run_experiments.sh

The run_experiments.sh script allows you to run experiments with different models and datasets. It also supports running experiments with multiple random seeds.

Usage ./run_experiments.sh

where <model> = {MISFEAT, MINE, MLP}
      <dataset> = {loan, mobile, attrition, sd1, sd2}

Computing Metrics

Script: compute_metrics.py After running experiments, you can compute metrics for the results using the compute_metrics.py script.

Usage python3 compute_metrics.py --dataset --model

where <dataset> = {loan, mobile, attrition, sd1, sd2}
      <model> = {MISFEAT, MINE, MLP}

Example Commands

1. Run MISFEAT on the loan dataset ./run_experiments.sh MISFEAT loan

2. Run MINE on the mobile dataset: ./run_experiments.sh MINE mobile

3. Run MLP on the attrition dataset: ./run_experiments.sh MLP attrition

4. Compute metrics for MISFEAT on mobile: python3 compute_metrics --dataset mobile --model MISFEAT