A Smart Air Pollution Detector Using Machine Learning

The rapid growth of urbanization and industrial activities in cities has resulted in a significant deterioration in air quality, which poses an increasing threat to both public health and the environment. This study focuses on predicting air quality using ML algorithms, aiming to classify air quality into three distinct categories: Good, Satisfactory, and Poor. The dataset utilized for this research comprises key environmental factors such as PM2.5, PM10, nitrogen oxides, and carbon monoxide, which are considered critical indicators of air pollution. To enhance the accuracy of predictions, several ML models were employed, including Logistic Regression, MLP, Random Forest, Decision Tree, The data preprocessing phase involved several essential steps to prepare the dataset for model training. These steps included the handling of missing values, selection of relevant features, and addressing class imbalance through the use of the SMOTE, which was employed to balance the distribution of target labels. The models were then trained and evaluated based on their performance in predicting air quality categories, with accuracy being the primary evaluation metric. Moreover, it can help inform public health decisions by identifying regions with poor air quality and ensuring better management of air pollution levels.