This project focuses on the indoor levels of the air pollutant PM2.5 in residential environments.
Two houses are used for this study with similar layouts.
Three rooms have been looked into: Kitchen, Living Room and Bedroom.
Each house has an observation per house, save some momentarily disconnections from the device, totalling to 131-139 observations per house.
Both houses have a joined Kitchen and Living room with only a half wall separating both.
With this project we aim at identifying the most important features to predict PM2.5 per room, investigate if those features differ between rooms (relationships between rooms), if they differ between houses (to establish if house properties are more predominant than the data we have collected (air pollutants through sensor features and occupant activity through questionnaires) and finally we'll try predicting PM2.5 per room through different models. The goal with the prediction is to see if we have enough data to predict PM2.5 with any degree of success, compare different models to see which one is the most appropriate for our data and compare models with and without our Occupant Activity to see just how impactful those are.