Summary: | 碩士 === 輔仁大學 === 統計資訊學系應用統計碩士在職專班 === 105 === Recently, the air pollution has been worsening. The PM2.5 is the particulate matter (PM) that has a diameter less than 2.5 micrometers. Because high level of PM2.5 can cause immediate health problems, the prediction of PM2.5 is important. This study uses the autoregressive integrated moving average (ARIMA), artificial neural networks (ANN), Support Vector Regression(SVR) and the integrated ARIMA-ANN, ARIMA-SVR approaches for predicting the PM2.5 in six municipalities of Taiwan. In this study, the forecasting accuracy measure is based on the mean absolute percentage error (MAPE). The practical dataset, from the years 2006 to 2015, for PM2.5 in six municipalities of Taiwan, are collected and analyzed. The PM2.5 prediction results report that the ARIMA-ANN model has the most satisfactory forecasting accuracy.
|