Hyperparameter optimization in regression model to predict atmospheric pollutants

In recent times, due to the increase in air pollution and its impact on the everyone’s health and environment; it has become a very tedious task to monitor and check the air quality. Predicting the air quality is a highly tangled task due to the volatility, particle nature and variables constituting the pollutants. It has become pertinent to check the air quality in rural/urban areas due to the impact it has on people’s health and the environment. In this paper, we first do comparative analysis of popular ML(machine learning) method, linear regression, random forest, XGBoost decision tree, k-nearest neighbors (KNN), and L1 and L2 regularization for forecasting pollutant and particulate levels and also the air quality index (AQI) is predicted. Then we apply cross validation and grid search to optimize our random forest model. Finally, we predict the model with Long short-term memory (LSTM).