This repository aims to implement sophisticated machine learning models using publicly available datasets in the bioinformatics and fintech domains. The sources and credits are duly mentioned in the respective projects, but key aids have been ProjectPro, Medium and Kaggle for additional datasets to augment the work done here. The projects implemented so far include:

linear regression on soccer data to predict the score a player may achieve in the English Premier League

image segmentation using UNET++ to detect the location of tumors in images from scans, spanned over many tissues like breast data, colorectal, hippocampus, spleen, heart and prostrate data.

The data itself is not uploaded but please reach out to me at https://www.linkedin.com/in/sthumsi/ for the same. I also have the trained model saved to file in a .h5 format. Training on only colorectal + breast data took 4 hours with 3 epochs, and 18 hours with the full dataset and 3 epochs.