Does PM10 influence the prediction of PM2.5?

Urbanization has led to a sharp increase in exposure to air pollutants in developing regions & the New Delhi capital of India is no exception to it. This paper proposes an approach where the Delhi region is divided into hexagonal bins of different sizes. Then the spatial interpolation is performed using Inverse distance weighting for pollutants and Ordinary Kriging for the meteorological parameters at the centroid of each bin. A hybrid deep learning architecture developed using convolutional neural network, and long short term memory is used for multivariate time series regression and prediction for PM2.5. Two different models are developed, one considering PM10 as a predictor variable and another without considering PM10. The results from both models are compared using various performance matrices, and experimental predicted results show that it improves prediction performance when PM10 is taken into account.