Open AccessMultiple Linear Regression (MLR) and Principal Component Regression (PCR) for Ozone (O3) Concentrations Prediction
Author(s) -
Nur Nazmi Liyana Mohd Napi,
Mohammad Syazwan Noor Mohamed,
Samsuri Abdullah,
Amalina Abu Mansor,
Ali Najah Ahmed,
Marzuki Ismail
Publication year - 2020
Publication title -
iop conference series. earth and environmental science
Language(s) - English
Resource type - Journals
eISSN - 1755-1307
pISSN - 1755-1315
DOI - 10.1088/1755-1315/616/1/012004
Subject(s) - multicollinearity , principal component regression , linear regression , principal component analysis , statistics , variance inflation factor , regression analysis , nitrogen dioxide , mathematics , regression , coefficient of determination , econometrics , meteorology , geography
Rapid economic growth has led to an increase in ozone (O 3 ) concentration which significantly affecting human health and environment. The prediction of O 3 is complicated due to the redundancy of influencing parameters which introduce the multicollinearity problem. The aim of this study is to assess the best prediction model for O 3 concentration which is Multiple Linear Regression (MLR) and Principle Component Regression (PCR). Data from 2012 to 2014 were used including O 3 , nitrogen dioxide (NO 2 ), nitrogen oxide (O 2 ), temperature, relative humidity and wind speed on hourly basis. Principle Component Analysis (PCA) was used in order to reduce multicollinearity problem, prior to the implementation of MLR. The hybrid model of PCR was selected as best -fitted models as it had higher correlation coefficient, R 2 values compared with MLR model. In conclusion, the information from best-fitted prediction model can be used by local authorities to plan the precaution measure in combating and preserve the better air quality level.