Selected for final offline round (top 10) among 2k+ registration and ~375 valid submissions . Couldn't attend the final round due to academic commitments .
Developed an ML Algorithm for the given datasets using physics guided feature engineering (catalytic reaction domain) together with ensemble learning achieving a RMSE of 10.94 , MAE of 2.76 and R2 score of 0.91.
Problem statement is publicly available and is attached in this repo taken from Fugacity ML Hackathon (IIT KGP 26) kaggle for reference . Please refer to the notebook for the end to end solution with clear decisions and explanations