My Projects
🏃 NextMove: Predicting Next Workout Activity
NextMove predicts a user’s next workout activity using historical fitness tracking data. It models sequential exercise behavior as a multiclass classification problem, leveraging feature engineering, heuristic baselines, and a Random Forest classifier to uncover temporal and physiological patterns in real-world Endomondo logs.
🔥 America on Fire: Mapping U.S. Fire Distribution
America on Fire is an interactive data visualization of U.S. wildfires in 2024 built with D3.js and NASA MODIS data, highlighting spatial patterns, seasonality, and unexpected hotspots.
🚴🏼♀️ Bikewatching
Bikewatching visualizes Boston and Cambridge bike activity with interactive maps, showing Bluebikes station traffic, trip patterns, and bike lane networks using Mapbox GL JS and D3.js.
🌊 ENSOcast: Decoding El Niño–Southern Oscillation
Presented at the San Diego Undergraduate Tech Conference 2025, ENSOcast is a climate prediction platform that decodes the El Niño–Southern Oscillation phenomenon using machine learning and decades of NOAA data. It leverages Random Forest, XGBoost, 1D CNN, LSTM, and ensemble models, achieving >80% accuracy in predicting monthly ENSO phases.
⚡ Predicting U.S. Power Outage Severity
Analyzing power outages in the continental U.S. (2000–2016) to predict outage duration and assess key contributing factors using data-driven approaches. Achieved 80% accuracy with Random Forest models.
Geriasphere
Geriasphere is a user-friendly website that empowers middle-aged and senior citizens in India to navigate essential mobile apps. Created using HTML and CSS, the website features step-by-step video tutorials for apps like Uber, Google Maps, Zoom, and Gmail, with a clean visual layout and app-specific navigation.