A Machine Learning Approach to Analyze the Noise Pollution Level of Schools of Rural and Urban Areas

Although it is a rural or urban environment, modern society sees a lot of traffic and crowded spaces everywhere. Because of this, global noise pollution is growing every day. Particularly in Bangladesh, a nation with a high population density, there is a lot of vehicular noise pollution. As a result, noise pollution disturbs people of all ages and causes several medical and psychological issues. Our goal in this project is to utilize machine learning and the Internet of Things (IoT) to construct a system for analyzing the degree of noise pollution. By developing and using an IoT-based system to detect noise pollution, we compile a primary dataset from various rural and urban schools, focusing on bus stops and roadside locations. We may categorize noise levels into three levels (Low, medium, and High) by measuring the noise level in decibels (dB). Following the collection of all the real-time data, we preprocessed the data. We used various machine learning methods, including Gradient Boosting, Random Forest, Support Vector Machine (SVM), and Naive Bayes. These algorithms provide high precision, recall, F1 score, and accuracy. Gradient Boosting provides 99% overall accuracy, 100% recall, 100% precision, and 100% f1-score rate. Our IoT-based noise monitoring system, which we implanted, along with the advanced machine learning techniques used to analyze its level, has the potential to significantly improve our overall health and the quality of our daily environment.