Kernel Ridge Regression

with gaussian kernel and k-Fold cross-validation

KRR

The five Matlab scripts found in the root directory of this repository are tools for using the kernel ridge regression algorithms. With the use of these matlab scripts you can easily implement and evaluate the KRR algorithm on any set of continuous floating point data.

Walkthrough

The /walkthrough directory contains resources that were designed to explain the inner workings of the algorithms written in this repository.

Gaussian Example

The /gaussian_example directory contains example demonstrating the usage of the KRR algorithms found in this repository.

Double Perovskites

The /double_perovskites directory contains an example implementation of the KRR algorithms from this repository. The goal is to learn to predict the bang-gaps of a material from it's chemical composition.

Data

The /data directory contains all the data files used in the examples. gaussian_data.csv represents a 3-dimensional function with added gaussian noise. double_perovskites_gap.xlsx contains chemical compositions and accompanying band-gaps of 1306 different double-perovskites.

The double-perovskites database was sourced from https://cmr.fysik.dtu.dk/

written by Robert Kuramshin