Integrating Ontology Semantics and Data-Driven Feature Pre-processing for Ice Breakup Prediction in the Canadian Rivers

• Tech Stack:Environmental Data Modeling, Ontology Semantics, Data-Driven Feature Pre-processing, Ice Breakup Prediction, Canadian Hydrology
• URL: Gitlab

• Implemented advanced feature selection methods, including SHAP and Permutation Importance, to refine feature rankings for better prediction accuracy.
• Developed and integrated five ensemble models (Borda Count, Condorcet, Coombs, Instant Runoff, Reciprocal Ranking) using XGBoost and Random Forest to predict mid-winter breakups in Canadian rivers.
• Collaborate with McMaster Uni (Canada)