index.md

title	layout
Preditive Modeling	base

Execute all the Notebook with

Machine Learning Models

Getting Started: Introduction
Installation, download data and extract segments and features for predictive analysis¶

A quick start with SVM (LWR task)Starting with LWR classification using Rhythmic features (spectral) without artifact removal

All four predictive tasks using SVM Applying ICA based artifact removal, Segment-wise feature extraction

Tuning the preprocessing parameters Tuning the parameters of filtering and Artifact removal method

Feature Extraction Framework Segment-wise & Window-wise feature extraction, for details check the article

Using External Libraries Extract EEG, GSR, and PPG signals, process with other libraries or your own custom function and apply predictive modeling

Using Decision Tree to analyse Applying decision tree to identify the important predictor from spectral features

Improve performance by using ATAR from spkit in phyaat to remove EEG Artifacts

EEG Features with custom frequency bands

Temporal Features with window-wise approach can be used with RNN/LSTM/Baysian Network

• With Wavelet Based artifact removal - will be updated soon

Deeplearning and Temporal Models

In this section, the exmples includes preditive modeling with convolutional and temporal models such as CNN, RNN and Bayesian Network

EEG to Image (SSFI)
Converting EEG Data to Spatio-Spectral Feature Image¶

Deep Representation of EEG
Using SSFI: Spatio-Spectral Feature Image and CNN¶

---

• Using 1D CNN - will be updated later
• Using LSTM - will be updated later

Contribute and record your results

Everyone is invited to use the dataset and build better models. You can record your performance metric on leaderboard at PaperwithCode here https://paperswithcode.com/dataset/phyaat

Additionally, you could also upload your code on cloud, to reproduce the results and once you do that, you could send us the link to your code and we will add your predictive model in above collections. In order to do that, make sure, your code is self-contained (includes all the dependencies and libraries) and easy to execute.