Peramalan PM2,5 Berdasarkan Indeks Kualitas Udara di DKI Jakarta Dengan Metode SETAR-Tree dan Long Short-Term Memory (LSTM)

Dinda, Ayu Safira (2025) Peramalan PM2,5 Berdasarkan Indeks Kualitas Udara di DKI Jakarta Dengan Metode SETAR-Tree dan Long Short-Term Memory (LSTM). Masters thesis, Institut Teknologi Sepuluh Nopember.

Text 6003231020-Master_Thesis.pdf - Accepted Version Restricted to Repository staff only until 1 April 2027. Download (2MB) | Request a copy

Abstract

Kualitas udara di DKI Jakarta mengalami penurunan yang signifikan akibat tingginya polusi udara, terutama polusi partikulat PM2,5, yang telah mengakibatkan dampak negatif bagi kesehatan manusia dan kerugian ekonomi yang besar. Partikulat PM2,5 memiliki kemampuan untuk melewati penyaringan rambut hidung dan mencapai saluran pernapasan. Salah satu cara untuk mengatasi masalah polusi udara adalah dengan menggunakan prediksi waktu berdasarkan data historis untuk meramalkan PM2,5 harian. Metode yang dilakukan pada penelitian ini yaitu model SETAR-Tree dan LSTM dalam meramalkan konsentrasi PM2,5 di DKI Jakarta. Data konsentrasi PM2,5 diambil dari stasiun pemantauan kualitas udara periode Januari 2022 hingga November 2023. Model SETAR-Tree menggabungkan prinsip model TAR dengan konsep pohon keputusan, sementara LSTM merupakan arsitektur jaringan saraf yang efisien dalam memprediksi deret waktu. Hasil pemodelan akan dibandingkan untuk mengevaluasi kinerja kedua model dalam memprediksi konsentrasi PM2,5. Berdasarkan hasil analisis yang dilakukan, pada pemodelan pada in sample dan out sample, SETAR-Tree memiliki performa yang lebih baik daripada LSTM karena nilai RMSE dan MAPE yang lebih kecil. Hasil peramalan menggunakan model LSTM, konsentrasi PM2,5 akan menurun nilainya sedangkan dari hasil peramalan yang diperoleh model SETAR-Tree, konsentrasi PM2,5 cenderung fluktuatif pergerakannya selama satu bulan kedepan
=====================================================================================================================================
Air quality in DKI Jakarta has decreased significantly due to high air pollution, especially PM2.5 particulate pollution, which has resulted in negative impacts on human health and major economic losses. PM2.5 particulates have the ability to pass through nasal hair filters and reach the respiratory tract. One way to overcome the problem of air pollution is to use time predictions based on historical data to forecast daily PM2.5. The methods used in this study are the SETAR-Tree and LSTM models in predicting PM2.5 concentrations in DKI Jakarta. PM2.5 concentration data was taken from air quality monitoring stations from January 2022 to November 2023. The SETAR-Tree model combines the principles of the TAR model with the concept of a decision tree, while LSTM is a neural network architecture that is efficient in predicting time series. The modeling results will be compared to evaluate the performance of the two models in predicting PM2.5 concentrations. Based on the results of the analysis carried out, in the modeling of in-sample and out-sample, SETAR-Tree has better performance than LSTM because the RMSE and MAPE values are smaller. The forecasting results using the LSTM model show that the PM2.5 concentration will decrease in value, while from the forecasting results obtained by the SETAR-Tree model, the PM2.5 concentration tends to fluctuate in its movement over the next month

Item Type:	Thesis (Masters)
Uncontrolled Keywords:	Air Quality, Long Short-Term Memory, PM2,5, SETAR, Time Series, Kualitas Udara, Long Short-Term Memory, PM2.5, SETAR, Time Series
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:	Dinda Ayu Safira
Date Deposited:	30 Jan 2025 01:51
Last Modified:	30 Jan 2025 01:51
URI:	http://repository.its.ac.id/id/eprint/116936

Actions (login required)

View Item