Kernel methods data challenge

Our attempt for the Kaggle data challenge Machine learning with kernel methods 2021 where we achieved 0.64600 accuracy and ranked 52th.

We implement various classification models. For our submission, we used the spectrum kernel with tuned hyperparameter k.

Usage

Requirements

• Python 3
• numpy

Installation

Clone the repository

git clone https://github.com/MB-29/kernel-challenge

and move to the root directory

cd kernel-challenge

Run prediction

Reproduce our submission by running the script predict_spectrum.py

python predict_spectrum.py path_to_data_directory