Three days ahead prediction of daily average concentration of PM10 using regression tree approach / Firhad Dhiyafique Harun, Wan Nur Shaziayani Wan Mohd Rosly and Prof Madya Dr Ahmad Zia Ul-Saufie Mohamad Japeri

The air pollution in Malaysia are always fluctuated throughout the year. This is corresponding to the growth of industrial area and emission from vehicle. These episodes had indirectly significant impact on the air quality of Malaysia. Hence this study aims to study the trend and predict the 3 days ahead of daily concentration of PM10 by using IBM SPSS and rapidMiner Studio respectively. The station is located at Jerantut, Pahang as it is significant to the centre of Peninsular Malaysia. As for the trend, air pollution hourly, monthly, and yearly monitoring records from 2004 until 2017 were used in analysing the statistical data analysis. There are eight parameters were selected in this study which known as PM10, CO, SO2, NO2, O3, relative humidity (RH), temperature (T) and wind speed (WS). The results obtained for the trend shows that it is higher reading of PM10 in June until September every year due to the factors are usually coming from the neighbouring country and wind direction projectile to Malaysia. For the prediction, the results obtained for Root Mean Squared Error are 10.164, 13.853, and 13.281 respectively for day 1, day 2 and day 3. As for absolute error, the results obtained are 5.893, 8.268, and 9.052 respectively for day 1, day 2 and day 3. The results also indicated that dispersion of PM10 in Malaysia were significantly affected by temperature, wind speed and relative humidity.