/seizures

A project that compares and contrasts classical and machine learning approaches to detecting seizures from EEG data and classifying EEG signals.

Primary LanguageJupyter Notebook

README

This repository contains a project that compares and contrasts the algorithmic / ML approach to data analysis with the classical statistical modeling approach.

It contains

  • an article that gives a detailed report on the topic;
  • the R code required to reproduce the classical statistical modeling results;
  • a Jupyter notebook for reproducing the ML results;
  • the original data for reproducing the analysis in full in the file ./data/seizures_original.csv;
  • three transformed data sets so that you can pick up at the modeling stage without having to re-run feature extraction;
    • the features extracted for the Python Jupyter notebook in the file ./data/seizures_features.csv;
    • the features extracted for the R code in the file ./data/fdata.csv;
    • autocovariance data produced as an intermediate feature extraction step (for the R file) in the file ./data/AC.csv.
  • finally, the repository contains the python package requirements in requirements.txt reproducibility.