Time Series–Based Forecasting of Ground-Level Ozone Concentration Using Machine Learning and Deep Learning Models

Ozone pollution poses serious environmental and health challenges, especially in urban regions. Accurate forecasting of ozone concentration levels enables early warning systems and supports policy-level decision-making. This final-year project focuses on ozone level forecasting using time series analysis techniques relevant to data analytics applications. Historical ozone concentration data were analyzed to identify trends, seasonality, and temporal dependencies. The Autoregressive Integrated Moving Average (ARIMA) model was implemented for prediction. Model performance was evaluated using Mean Absolute Error (MAE) and Root Mean Square Error (RMSE). Results demonstrate that time series models are effective for short-term ozone forecasting and are suitable for real-world environmental analytics applications.