Page 77 - programme book

ST-007
An Efficient Tool for Monitoring Air Quality (PM10) at Several Cities in
Peninsular Malaysia Using Functional Principal Component Control Chart
with Data Pre-Whitening

Norshahida Shaadan 1, a) , Abdul Aziz Jemain 2, b) and Zamira Hasanah Zamzuri 2, c)

1 Center of Statistics and Decision Science, Faculty of Computer and Mathematics, Universiti Teknologi MARA,
40450, Selangor, Malaysia
2 School of Mathematical Sciences, Faculty of Science & Technology, Universiti Kebangsaan Malaysia (UKM),
43600, Bangi, Selangor, Malaysia

a) Corresponding author: shahida@tmsk.uitm.edu.my
b) azizj@ukm.edu.my
c) zamira@ukm.edu.my

Abstract. The use of a control chart to monitor air quality performance is important to assist in the
management of the air pollution problem. This study aims to propose a control chart that enables the
monitoring of daily air quality (PM10) over time. The control chart is developed based on daily curves
data using Functional Principal Component Analysis. Using a control chart, any deterioration of air
quality (PM10) can be assessed based on the pattern of detected anomalies. Continuous monitoring
can be conducted using daily air quality indices, which are developed using the Hotelling T statistics.
2
To reduce the autocorrelation effect that may cause false alarm, data- pre-whitening is employed in
this study. The application of the control chart was conducted using historical data, involving a seven-
year period of data starting from 2004-2010 at several urban air quality monitoring stations located in
Peninsular Malaysia. The results of the analysis have shown that this functional data-based control
chart outperforms the classical average-based control chart in terms of the number of indecisive out-
of-control points. The implementation of data pre-whitening approach also reduces false alarms. In the
context of an application, the results also show that the cities located in the west part of Peninsular
Malaysia are more likely to experience more abnormal air quality (PM10). The control chart is proven
to be practically useful as a monitoring tool to easily visualize the normal and abnormal trend of daily
air quality (PM10) in this study.

Keywords: Air Quality, Control Chart, Functional Data Analysis, Data pre-whitening, Statistical
process control

75