Machine Learning Based Prediction of PM 2.5 Pollution Level in Delhi

Scrutinizing Air pollution stances challenges due to the huge quantity of alignments present in the Air. Predicting PM 2.5 levels allows for further analysis and prediction of quality of air. PM 2.5 forms a major component of air pollution. This work addresses various machine learning algorithms to predict levels of PM 2.5, which are abundant in the atmosphere. We transformed problem into a binary classification with two classes being moderate and polluted. Support vector machine, Naïve Bayes, K-nearest neighbors, random forest algorithms, and Principal component analysis (PCA) were applied to obtain results. The prediction scores are favorable with support vector classification kernel giving the best result. Results from random forest and Naïve Bayes are similar while Naïve Bayes having a much lower predicting accuracy. PCA approach does not hold much significance as it gives a much lower prediction score.