IoT and Machine Learning Based Air Quality Monitoring and AQI Prediction System

Abstract: Air pollution has emerged as one of the most critical environmental and public health challenges of the twenty- first century. In India, the majority of urban centres regularly report Air Quality Index (AQI) values in the 'Poor' to 'Hazardous' range, yet real-time, localised air quality data remains scarce and inaccessible due to the high cost and sparse deployment of certified monitoring stations. This paper presents an end-to-end IoT and Machine Learning Based Air Quality Monitoring and AQI Prediction System that addresses these limitations through affordable hardware, cloud connectivity, and intelligent data analysis. The hardware sensing node is built around an ESP32 microcontroller interfaced with five sensors: a DHT11 temperature and humidity sensor, an MQ2 smoke and combustible gas sensor, an MQ7 carbon monoxide sensor, an MQ135 air quality gas sensor, and an optical PM2.5 dust sensor. Readings are displayed locally on a 20×4 I2C LCD and uploaded to a ThingSpeak cloud channel every 15 seconds. The machine learning subsystem employs a Random Forest classifier trained on 1,000 labelled environmental records spanning six AQI categories. The trained model is deployed within a Streamlit web application supporting Manual Input and ThingSpeak Auto modes, generating an AI Environment Report for every prediction. System validation used 82 live readings collected from the prototype hardware on 31 March 2026. The proposed system demonstrates that integration of low-cost IoT sensing, cloud data management, and ensemble machine learning can produce an intelligent air quality monitoring platform suitable for educational institutions and smart city deployment.

Keywords: Internet of Things, Air Quality Index, Arduino UNO, ESP32, MQ Gas Sensors, PM2.5, DHT11, ThingSpeak, Random Forest, Machine Learning, Streamlit, Environmental Monitoring, AQI Prediction, Smart Environment.