water-quality-modeling

Science and technical description of the Cockburn Sound Model

Lab for Innovation Science at Harvard data science competition. My solution scored around the top 15. I had planned to try other things, but I used only 1/3 of the allowed competition time -- and hence did not explore advanced feature engineering in the winning solutions. But interestingly, this solution was similar to winning solutions.

A machine learning system that predicts Water Quality Index (WQI) and Water Quality Classification (WQC) using engineered water-quality features. The project includes full preprocessing with missing-value handling, outlier treatment, dynamic WQI calculation, and both regression and classification modeling powered by XGBoost and Scikit-learn.

Browser tool for water-quality analysis, WHO/EU limits, and treatment support.