HARNESSING MACHINE LEARNING FOR ACCURATE AIR QUALITY PREDICTION: A TIM E SERIES ANALYSIS

Abstract

The existence of dangerous or excessive amounts of pollutants in the atmosphere that endanger human health, the environment, and general quality of life is referred to as air pollution. Short term exposure to air pollution results in acute health issues such headaches, exhaustion, and irritation of the throat, nose, and eyes. Prolonged exposure raises the risk of heart disease, lung cancer, chronic respiratory disorders, brain, nerve, liver, and kidney damage. Children who are exposed for an extended period of time may also have problems. As a result, assessing and forecasting air quality is a crucial step in reducing environmental risk. To measure pollution levels, we use AQI. In recent years, there has been a growing interest in the use of machine learning (ML) and deep learning (DL) techniques for air pollution forecasting. These techniques have the potential to provide more accurate and timely predictions of air pollution levels, which can be used to inform public health interventions and environmental policy decisions. This thesis reviews the existing literature on the use of ML and DL techniques for air pollution forecasting. The thesis provides an overview of the different types of ML algorithms that have been used for this purpose. For the previous two years, Delhi, the capital of India, has been the most polluted city in the world. So, this research paper collected data on air pollution from CPCB (Central Pollution Control Board) specifically focusing on fine particles such as PM2.5, PM10, NO2, NH3, SO2, CO and OZONE from five different areas or stations in Delhi. Different machine learning algorithms such as Logistic Regression, K Nearest Neighbors (KNN), Support Vector Machine (SVM), Random Forest, Decision Tree and XGBoost are used to analyses the collected data. Evaluation metrics such as accuracy, Precision and are used. The comparison between the models is also discussed in this thesis.