Over the years, the air quality in the capital city of Delhi and its adjoining states has reached an alarming state, primarily due to the increasing concentration of air pollutants such as PM2.5. This heightened concentration has been linked to severe respiratory problems. In this study, we assess the forecasting of PM2.5 concentration specifically in the Gurugram region of Haryana. We employ time series analysis and various regression models on weather data obtained from the Indian Meteorological Department (IMD) to construct a prediction model on an hourly basis. The study involves a comprehensive comparison of different models, encompassing data processing, exploratory data analysis (EDA), model development, and the application of performance metrics such as Root Mean Squared Error (RMSE) and R2 Score. The objective is to evaluate the error, which represents the difference between actual and predicted values. Upon comparing various time series and regression models, it is observed that the Long Short-Term Memory (LSTM) based analysis outperformed Adaboost and other regression models, exhibiting lower performance metrics. This suggests that the LSTM model provides a more accurate prediction of PM2.5 concentration in the specified region and timeframe.