Analysis of statistical models for forecasting PM10 in Kototabang region

Abstract

PM10 is one of the aerosol particles that can endanger human health. This research conducted by forcasteing for PM10 concentration. Forecasting is an activity of estimating or predicting events in the future, therefore it is necessary to do analysis simple statistical model to know goog results. In this case, several forecasting models are used for the daily PM10 concentration in Kototabang, that is Liniear, Quadratic, and Exponential Trend Model. As the results of this research, monthly forecasting using Linear Trend Model has the highest correlation value in October (-60) with MAD value 0.254, MSE 0.0651, RMSE 0.255, and MAPE 1848.8. Monthly forecasting using the Quadratic Trend Model has the highest correlation value in February (+0.52) with MAD value 0.013, MSE 0.0002, RMSE 0.015, and MAPE 103.934. then, for monthly forecasting using the Exponential Trend Model has the highest value in October (+0.61) with MAD value 0.124, MSE 0.0154, RMSE 0.124, and MAPE 893.484. the output of the forecasting model obtained the best model for forecasting the daily PM10 concentration in Kototabang i.e by using a simple statistical model, linear trend model.