The Comparison of PM2.5 forecasting methods in the form of multivariate and univariate time series based on Support Vector Machine and Genetic Algorithm

This research aims to study and compare the forecasting precision of multivariate and univariate time series forecasting based on Support Vector Machine optimized with Genetic Algorithm. A study data is an hourly data set contains the PM2.5 data and a meteorological data in Beijing, China. This data is published in UCI Machine Learning Repository. This study starts with a generating of 3 data subsets from an original study data. Each data subset will be used to generate a multivariate and univariate time series model to forecast PM2.5. After that, we evaluate our research with error measurement by using RMSE and MAPE. From the results, we found that univariate time series models have lower error than multivariate time series models for all of 3 data subsets.