Prediksi Indeks Standar Pencemar Udara Di Provinsi DKI Jakarta Menggunakan Pendekatan Vector Autoregressive-Based Long Short Term Memory

Ningrum, Ariska Fitriyana (2023) Prediksi Indeks Standar Pencemar Udara Di Provinsi DKI Jakarta Menggunakan Pendekatan Vector Autoregressive-Based Long Short Term Memory. Masters thesis, Institut Teknologi Sepuluh Nopember.

Text 6003202004-Master_Thesis.pdf - Accepted Version Restricted to Repository staff only until 1 April 2025. Download (2MB) | Request a copy

Abstract

Pemantauan ISPU dilakukan berdasarkan data meteorologi yang mempengaruhi konsentrasi udara ambien. Data yang dimaksud seperti kecepatan dan arah angin, kelembaban, temperature udara, intensitas matahari, dan curah hujan.Pada web SILIKA DKI Jakarta belum ada fitur untuk melihat prediksi ISPU padahal fitur tersebut bermanfaat bagi masyarakat. Keuntungan dari memprediksi ISPU adalah masyarakat dapat mengantisipasi lebih awal terkait kondisi kualitas udara yang akan terjadi termasuk pencemaran udara. Dalam studi time series, dilakukan prediksi multivariate time series terhadap parameter ISPU yang saling berkorelasi. Metode konvensional seperti Vector Autoregressive. Model VAR merupakan pengembangan dari model Autoregressive (AR) dimana pada model VAR variabel endogen yang digunakan lebih dari satu. Selain itu, dalam penelitian ini digunakan pula metode peramalan Artificial Intelligence seperti Long Short-Term Memory yang dapat mengatasi non-linieritas pada data time series. Hasil estimasi parameter pada LSTM mengunakan propagation error dengan gradient decent untuk mendapatkan nilai bobot W dan U pada setiap gate. Hasil prediksi Konsentrasi PM2,5 dan PM10, menggunakan Vector Autoregressive dengan lag optimum adalah 5 diperoleh variabel yang signifikan pada PM2,5 dipengaruhi oleh variabel PM2,5 pada t-1 sebelumnya, sedangkan pada PM10 dipengaruhi oleh t-1 pada variabel PM2,5 dan PM10 pada t-5 sebelumnya. Berdasarkan hasil perbandingan kedua model menggunakan nilai RMSE dan MAPE didapatkan model terbaik pada prediksi parameter ISPU PM2,5 dan PM10 adalah menggunakan metode LSTM dengan nilai RMSE dan MAPE terkecil yaitu 13,69 dan 0,14 pada variabel PM2,5 dan 9,02 dan 0,14 untuk nilai RMSE dan MAPE pada variabel PM10
===================================================================================================================================
API monitoring is carried out based on meteorological data affecting ambient air concentrations. The data referred to include wind speed and direction, humidity, air temperature, solar intensity, and rainfall. On the SILIKA DKI Jakarta website there is no feature to see API predictions even though this feature is useful for the community. The advantage of predicting API is that the community can anticipate earlier related to air quality conditions that will occur including air pollution. In a time series study, multivariate time series predictions were made for API parameters that were correlated with each other. Conventional methods such as Vector Autoregressive. The VAR model is a development of the Autoregressive (AR) model where in the VAR model more than one endogenous variable is used. In addition, this research also uses Artificial Intelligence forecasting methods such as Long Short-Term Memory which can overcome non-linearity in time series data. The parameter estimation results in the LSTM use a propagation error with a decent gradient to get the W and U weight values for each gate. The predicted results of PM2,5 and PM10 concentrations, using the Autoregressive Vector with an optimum lag of 5, obtained a significant variable in PM2,5 which is affected by the PM2,5 variable in the previous t-1, while in PM10 it is affected by t-1 in the PM2,5 variable. and PM10 on the previous t-5. Based on the results of a comparison of the two models using RMSE and MAPE values, it was found that the best model for predicting PM2,5 and PM10 ISPU parameters was using the LSTM method with the smallest RMSE and MAPE values, namely 13.69 and 0.14 in PM2,5 and 9.02 and 0.14 for the RMSE and MAPE values in the PM10 variable

Item Type:	Thesis (Masters)
Uncontrolled Keywords:	Indeks Standar Pencemar Udara, Vector Autoregressive, Long Short Term Memory, Air Pollutant Index, Vector Autoregressive, Long Short Term Memory
Subjects:	G Geography. Anthropology. Recreation > GE Environmental Sciences Q Science > QA Mathematics > QA276 Mathematical statistics. Time-series analysis. Failure time data analysis. Survival analysis (Biometry) Q Science > QA Mathematics > QA336 Artificial Intelligence Q Science > QA Mathematics > QA76.87 Neural networks (Computer Science)
Divisions:	Faculty of Science and Data Analytics (SCIENTICS) > Statistics > 49101-(S2) Master Thesis
Depositing User:	ARISKA FITRIYANA NINGRUM
Date Deposited:	20 Feb 2023 02:47
Last Modified:	20 Feb 2023 02:47
URI:	http://repository.its.ac.id/id/eprint/97616

Actions (login required)

View Item