Brief

This project, created primarily for “dognosis” has had me invested quite a bit, I’ve been studying things that had always deemed “not easy at all. All this snooping around has been beneficial, I believe, since I went through the entire process of making a pipeline for the current task, and I now have a lot more insight on what tweaks can be made, what practices should be used, and so on.

The several aspects involved are:

• Dataset Building
• Feature Extraction
• Machine Learning and Models

Colab Notebook:

The notebook can be found in the associated GitHub Repository.

If a reviewer wishes to run this entire notebook, please set up the following directory in google drive:

1. Create a folder named "MNE-Dognosis" in your google drive.
2. Place the following files from the dataset folder directly in this folder:
   • pilot_eeg.cdt.cnt
   • experiment_data_segment_1
   • experiment_data_segment_2

The following document contains a few of the visualization results in the process, but I recommend going through the code and running it, since the graphs are interactive.

Visualisations

Dataset & Data Analysis

The current provided dataset gave me an insight on how to handle EEG data, but it does contain several points from which we can learn, and focus on building a better, more robust dataset, that remains relevant even as the project progresses.

A few preprocessing techniques like Band filtering, ICA, have been used in the notebook as well.

The details on building a dataset for future use are in the document below: