IoT-Enabled Dataset Development for Air Pollution Prediction Using Machine Learning

Air pollution continues to pose a serious threat both to the environment and public health in Bangladesh’s fast-growing cities such as Dhaka, Gazipur, and Narayanganj. Traditional air quality monitoring networks are not only expensive but also lack spatial coverage. To address these issues, this study introduces a novel air quality monitoring system based on the Internet of Things (IoT) and machine learning (ML) for predicting the Air Quality Index (AQI). The system employs low-cost sensors like the DHT22 sensor for temperature and humidity, MQ series for gas detection, and PMS5003 for particulate matter. The sensors are connected to ESP8266 and Arduino Nano microcontrollers. Data gathered from various urban and industrial regions is uploaded to Firebase for storage and further analysis. The merged dataset was uploaded on Kaggle, where some of the ML algorithms such as Linear Regression, Support Vector Machine (SVM), Artificial Neural Network (ANN), K-Nearest Neighbors (KNN), Random Forest, and XGBoost were tried. After preprocessing and balancing the data using the SMOTE method, the data were ranked in terms of their predictive accuracy among the algorithms. XGBoost emerged with the highest predictive accuracy (R2= 1.0000, RMSE = 0.0003), which indicates the strength of ensemble learning towards AQI prediction. These findings show that the proposed IoT-ML system offers a dependable, scalable, and affordable solution for real-time air quality monitoring in Bangladesh.