Hybrid Tsr-gstarma Untuk Meramalkan Kualitas Udara di Gresik, Surabaya dan Tuban

Riskianto, Riskianto (2026) Hybrid Tsr-gstarma Untuk Meramalkan Kualitas Udara di Gresik, Surabaya dan Tuban. Masters thesis, Institut Teknologi Sepuluh Nopember.

Text 6003222013_Master_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (2MB) | Request a copy

Abstract

Air Quality Index (AQI) merupakan indikator kualitas udara yang menggambarkan tingkat konsentrasi polutan udara yang diklasifikasikan ke dalam beberapa kategori beserta risiko kesehatannya. Particulate Matter berukuran 2,5 μg⁄m^3 (PM2.5) merupakan salah satu polutan utama yang digunakan dalam penentuan AQI karena memiliki dampak signifikan terhadap kesehatan manusia serta ketersediaan data yang relatif luas. Mengingat masih terbatasnya penelitian terkait pemodelan dan peramalan kualitas udara di Provinsi Jawa Timur, penelitian ini bertujuan untuk menerapkan model Generalized Space–Time Autoregressive Moving Average (GSTARMA) dan model Hybrid Time Series Regression–GSTARMA (TSR–GSTARMA) dalam meramalkan konsentrasi PM2.5. Data yang digunakan merupakan data konsentrasi PM2.5 dari tiga lokasi yaitu Gresik, Surabaya dan Tuban. Kinerja model dievaluasi menggunakan Root Mean Square Error (RMSE) dan Symmetric Mean Absolute Percentage Error (sMAPE) untuk mengakomodasi perbedaan skala data antar lokasi. Hasil penelitian menunjukkan bahwa model Hybrid TSR–GSTAR(1) dengan bobot invers jarak memberikan performa peramalan terbaik secara keseluruhan dengan nilai RMSE in-sample masing-masing sebesar 5,020 pada Gresik, 6,871 pada Surabaya, dan 3,421 pada Tuban, serta rata-rata sMAPE sebesar 17,60. Pada data out-sample model Hybrid TSR–GSTARMA menghasilkan RMSE sebesar 3,857 pada Gresik, 7,797 pada Surabaya, dan 14,067 pada Tuban, dengan rata-rata sMAPE sebesar 40,56. Meskipun nilai error pada lokasi Tuban relatif tinggi pada data out-sample yang diduga disebabkan oleh gangguan alat pengukuran PM2.5. Hasil ini mengindikasikan bahwa model Hybrid TSR–GSTARMA memiliki potensi yang kuat sebagai pendekatan peramalan kualitas udara berbasis spasial-temporal di Provinsi Jawa Timur.
==================================================================================================================================
The Air Quality Index (AQI) is an indicator of air quality that represents the concentration levels of ambient air pollutants, which are classified into several categories along with their associated health risks. Particulate Matter with an aerodynamic diameter of 2.5 μg/m³ (PM2.5) is one of the primary pollutants used in AQI determination due to its significant impact on human health and the relatively wide availability of its data. Given the limited number of studies focusing on air quality modeling and forecasting in East Java Province, this study aims to apply the Generalized Space–Time Autoregressive Moving Average (GSTARMA) model and the Hybrid Time Series Regression–GSTARMA (TSR–GSTARMA) model to forecast PM2.5 concentrations. The data used consist of PM2.5 concentration measurements from three locations, namely Gresik, Surabaya, and Tuban. Model performance is evaluated using the Root Mean Square Error (RMSE) and the Symmetric Mean Absolute Percentage Error (sMAPE) to accommodate differences in data scales across locations. The results indicate that the Hybrid TSR–GSTAR(1) model with inverse distance weighting provides the best overall forecasting performance, with in-sample RMSE values of 5.020 for Gresik, 6.871 for Surabaya, and 3.421 for Tuban, and an average in-sample sMAPE of 17.60. For the out-of-sample data, the Hybrid TSR–GSTARMA model yields RMSE values of 3.857 for Gresik, 7.797 for Surabaya, and 14.067 for Tuban, with an average sMAPE of 40.56. Although relatively high error values are observed at the Tuban location in the out-of-sample period, which are presumed to be caused by malfunctions in the PM2.5 monitoring instrument, the overall results indicate that the Hybrid TSR–GSTARMA model has strong potential as a spatial–temporal air quality forecasting approach in East Java Province.

Item Type:	Thesis (Masters)
Uncontrolled Keywords:	Hybrid, Spasial-Temporal, PM2.5, GSTARMA, TSR
Subjects:	Q Science > QA Mathematics > QA276 Mathematical statistics. Time-series analysis. Failure time data analysis. Survival analysis (Biometry)
Divisions:	Faculty of Science and Data Analytics (SCIENTICS) > Statistics > 49101-(S2) Master Thesis
Depositing User:	Riskianto Riskianto
Date Deposited:	12 Feb 2026 03:55
Last Modified:	12 Feb 2026 03:55
URI:	http://repository.its.ac.id/id/eprint/132390

Actions (login required)

View Item