MAQMS — Air Quality Prediction
A real IoT sensor network and LSTM prediction pipeline built for my capstone at IPN — later extended into a peer-reviewed publication. This page runs on the same real sensor data.
View full source & documentation on GitHub →Architecture
Click a stage to see how it actually works — real MQTT topics, real Stream Analytics query, real model code.
Stations
Four real stations, real historical readings. Click one to plot its PM2.5/PM10 history.
Real training results
Actual output from the training notebook — Evelina's PM2.5 model. Blue is training data, orange is the held-out test set, green is the model's prediction.
