HFMOEA

This is the official implementation of our paper titled "HFMOEA: A Hybrid Framework for Multi-objective Feature Selection" published in Journal of Computational Design and Engineering, Oxford.

Requirements

To install the required libraries run the following in the command prompt:
pip install -r requirements.txt

Using the HFMOEA Algorithm

Run the following code in Command Prompt:
python main.py --path path/to/file/csv_name.csv

By default it is assumed that the csv file contains no headers. But if it does, then add the argument --csv_header True in the code above, otherwise an error will be triggered. All csv files must have the class labels in the last column of the file as integer/float values.

Other available arguments are listed as follows:

• popsize: Population Size (Note: must be equal to or more than 10, since 10 filter methods are used to initialize part of the population- refer to our paper for more details)
• generations: Number of generations for the HFMOEA algorithm
• mutation: Percentage of mutation
• topk: "topk" number of features (Please refer to our paper for more details)
• save_fig: Whether the plots are to be saved or not

Citation

If you find our repository useful, please consider citing our paper:

@article{kundu2022hfmoea,
  title={HFMOEA: A Hybrid Framework for Multi-objective Feature Selection,
  author={Kundu, Rohit and Mallipeddi, Rammohan},
  journal={Journal of Computational Design and Engineering},
  year={2022},
  publisher={Oxford University Press}
}