Indoor Air Quality Prediction Using Machine Learning Techniques

Monitoring indoor environment parameters became one of a required service nowadays. Because of the higher consumption of fossil fuel and less depending on renewable energy, climate is changing into a negative and dangerous direction and caused climate overheating as well as increased air pollution rates to extreme levels. From here on, monitoring the indoor air quality and having a kind of evaluation system to predict the air quality in advance is required and can provide a great added value for people living in closed environments and under strong dusty, polluted and highly toxic weather conditions. In this research we provide a mixed based solution between indoor air environment measurements system and machine learning based solution to predict and evaluate the current indoor air quality. Several ML models like Support Vector Machine (SVM) and Artificial Neural Networks (ANN), Random Forest, etc are used to implement the machine learning models and a simple hardware implementation is designed using AT - Mega microcontroller attached with Raspberry - Pi 4 to finalize indoor parameters measurements and machine learning processes. Temperature, Humidity, Dust and Gas sensors were used to measure indoor environment parameters. Our system could reach a classification accuracy of 100% using RF, 96% using SVM and 99% using ANN.