Zarah berdiameter aerodinamik kurang daripada 10μm (PM10) adalah salah satu daripada udara yang boleh menjejaskan kesihatan manusia. Tujuan kajian ini adalah untuk meramal kepekatan zarah untuk hari esok (PM10D1) dengan menggunakan Model Regresi Linear Berganda (MLR) dan Model Regresi Pokok Penggalak (BRT). Data min setiap hari digunakan dari 2013 hingga 2015 dibahagikan kepada data latihan (70%) dan data pengesahan (30%). Parameter yang mempengaruhi kepekatan PM10 untuk hari seterusnya adalah zarah terampai (PM10D0), kelajuan angin (WS), suhu (T), kelembapan relatif (RH), sulfur dioksida (SO2), nitrogen dioksida (NO2), ozon (O3) dan karbon monoksida (CO). Data min harian telah dipilih di empat stesen pemantauan iaitu Jerantut (stesen latar belakang), Nilai (kawasan perindustrian), Seberang Jaya (kawasan sub-bandar) dan Shah Alam (kawasan bandar). Keputusan yang diperolehi menunjukkan bahawa setesen Nilai merekodkan nilai kepekatan PM10 tertinggi berbanding stesen-stesen lain. Sumbangan utama pencemar udara di stesen Nilai adalah zarah terampai (PM10D0), karbon monoksida, nitrogen dioksida dan ozon. Keputusan yang diperolehi menunjukkan bahawa model Regresi Linear Berganda (MLR) adalah model yang terbaik untuk meramal kepekatan PM10 hari seterusnya berbanding model Regresi Pokok Penggalak (BRT).
_______________________________________________________________________________________________________
Particulate matter with an aerodynamic diameter less than 10μm (PM10) is one of the pollutants that can adversely affect human health. The aims of this study is to predict particulate matter concentration for the next day (PM10D1) by using Multiple Linear Regression (MLR) and Boosted Regression Trees (BRT) models. The daily mean data used from 2013 until 2015 is divided into training data (70%) and validation data (30%). The parameters that influence PM10 concentration for the next day are particulate matter (PM10D0), wind speed (WS), temperature (T), relative humidity (RH), sulphur dioxide (SO2), nitrogen dioxide (NO2), ozone (O3) and carbon monoxide (CO). Daily mean data were selected at four monitoring stations which are Jerantut (background station), Nilai (industrial area), Seberang Jaya (sub-urban area) and Shah Alam (urban area). The results obtained shows that Nilai station recorded the highest mean value of PM10 concentration compared to other stations. The main contributions of air pollution at Nilai station are particulate matter (PM10D0), carbon monoxide, nitrogen dioxide and ozone. The result shows that Multiple Linear Regression models (MLR) is the better model to predict the next day of PM10 concentration compared to Boosted Regression Trees (BRT).
Prediction of PM10 using multiple linear regression and boosted regression trees / Nur Haziqah Mohd Hamid