Forecasting Air Pollution Index in Klang by Markov Chain Model

Abstract

The main purpose of analyze future air quality is to maintain the environment in good and healthy condition. Current techniques applied to forecast the air pollution index were ARIMA, SARIMA, Artificial Neural Network, Fuzzy Time Series, Machine Learning, etc. Thus, each technique has its own advantages and disadvantages in the variables, model selection and model accuracy determination. This study aims to forecast air pollution index by developing a Markov Chain model in Klang district, Selangor state which is one of the most polluted area in Malaysia. The Markov Chain model development is a stochastic process sequence that depends on the previous successive event in time. In this model development, state transition matrix and probability are the main concept in determine the future behavior of Air Pollution Index which depends on the present state of the process. The result shows that the developed model is a simple and good performance model that will forecast and evaluate the distribution of the pollution level in long term.

Estimasi Bayes Empirik pada Model Rantai Markov untuk Menggambarkan Karakteristik Curah Hujan di Kota Makassar

“…Selanjutnya Sanusi [6] juga telah menggunakan model rantai Markov homogen orde pertama untuk menganalisis karakteristik perilaku kekeringan di Semenanjung Malaysia berdasarkan Indeks Presipitasi Terstandarisasi. Selain penggunaannya dalam data curah hujan, model rantai Markov juga telah digunakan untuk mengetahui dan memprediksi polusi udara di Klang Malaysia [7]. Model rantai Markov juga telah digunakan untuk menganalisis tingkat kenyamanan di kota Majene berdasarkan Temperature Humidity Index (THI) [8].…”

“…c. Mengestimasi nilai durasi keadaan kering, lembab, dan basah menggunakan Persamaan (6). d. Mengestimasi nilai periode ulang keadaan kering, lembab dan basah menggunakan Persamaan(7). e. Menginterpretasikan hasil yang diperoleh.…”