Extrapolative Data Feature Exploration-Based Air Quality Prediction Using Gradient Descent Algorithm

With a heightened awareness of the critical impact of air quality on public health and daily life, effective air quality prediction has emerged as a pivotal research area. This study delves into the intricate task of predicting the Air Quality Index (AQI) for various states in India, navigating challenges arising from the dynamic nature of data sources and pollutant concentration fluctuations over time. The predictive modeling strategy adopts a combination of linear regression and gradient descent techniques, tailored to the specifics of the AQI estimation task. A meticulous data preprocessing phase involves the identification and removal of outliers through Boxplot analysis, ensuring the robustness and reliability of the predictive models. To refine prediction accuracy, the study meticulously examines temporal variations in AQI. This exploration involves employing statistical tools to discern patterns, unraveling monthly fluctuations in AQI across different months. The resulting insights provide a nuanced understanding of how air quality varies throughout the year. Furthermore, the research extends its predictive capabilities to anticipate AQI values for specific months, focusing notably on January. Leveraging linear regression, the model’s effectiveness is rigorously evaluated using root mean squared error calculations. Visual representations aid in conveying the nuances of the predictive accuracy and model performance. In a forward-looking perspective, the study embarks on forecasting future AQI values. Utilizing gradient descent and linear regression, the model predicts AQI for upcoming years, offering a glimpse into potential air quality scenarios. The accuracy of these predictive models is systematically assessed, contributing valuable insights for long-term air quality planning and management.