Time Series based Air Pollution Forecasting using SARIMA and Prophet Model

Air pollution severely affects many countries around the world causing serious health effects or death. Increasing dependency on fossil fuels through the last century has been responsible for the degradation in our atmospheric condition. Pollution emitting from various vehicles also cause an immense amount of pollution. Pollutants like RSPM, SO2, NO2, SPM, etc. are the major contributors to air pollution which can lead to acute and chronic effects on human health. The research focus of this paper is to identify the usefulness of analytics models to build a system that is capable of giving a rough estimate of the future levels of pollution within a considerable confidence interval. Rendered linear regression techniques are found to be insufficient for the time-dependent data. In this regard, we have used time series forecasting approach for predicting the future levels of various pollutants within a considerable confidence interval. The experimental analysis of the forecasting for the air pollution levels of Bhubaneswar City indicates the effectiveness of our proposed method using SARIMA and Prophet model.