AIR QUALITY PREDICTION IN JAKARTA USING LSTM AND GRU MODELS FOR INFORMATION DISSEMINATION

Nathalia, Ni Putu Nita (2026) AIR QUALITY PREDICTION IN JAKARTA USING LSTM AND GRU MODELS FOR INFORMATION DISSEMINATION. Undergraduate thesis, Universitas Pendidikan Ganesha.

[img] Text (COVER)
2215101050-COVER.pdf
Download (757kB)
[img] Text (ABSTRAK)
2215101050-ABSTRAK.pdf
Download (195kB)
[img] Text (BAB 1 PENDAHULUAN)
2215101050-BAB 1 PENDAHULUAN.pdf
Download (11MB)
[img] Text (BAB 2 KAJIAN TEORI)
2215101050-BAB 2 KAJIAN TEORI.pdf
Restricted to Repository staff only
Download (11MB) | Request a copy
[img] Text (BAB 3 METODOLOGI PENELITIAN)
2215101050-BAB 3 METODELOGI PENELITIAN.pdf
Restricted to Repository staff only
Download (11MB) | Request a copy
[img] Text (BAB 4 HASIL DAN PEMBAHASAN)
2215101050-BAB 4 HASIL DAN PEMBAHASAN.pdf
Restricted to Repository staff only
Download (11MB) | Request a copy
[img] Text (BAB 5 PENUTUP)
2215101050-BAB 5 PENUTUP.pdf
Restricted to Repository staff only
Download (11MB) | Request a copy
[img] Text (DAFTAR PUSTAKA)
2215101050-DAFTAR PUSTAKA.pdf
Download (11MB)
[img] Text (LAMPIRAN)
2215101050-LAMPIRAN.pdf
Download (11MB)

Abstract

Recent time air quality has deteriorated due to the rise in industrial activities, transportation, and urbanization particularly in DKI Jakarta. This study aims to understand the performance of each model in predicting air quality and identify the most suitable model. The deep learning models employed in this research are Long Short-Term Memory (LSTM) and Gated Recurrent Unit (GRU). The methodology follows the CRISP-DM framework. However, this study is limited to the Evaluation stage.The dataset used covers a one-year period and was obtained from five selected monitoring stations: DKI1-Bundaran HI, DKI2-Kelapa Gading, DKI3-Jagakarsa, DKI4-Lubang Buaya, and DKI5-Kebon Jeruk. The training and validation processes were carried out using Time Series Cross-Validation (k=5). The model architecture comprises two hidden layers, two dropout layers, and one input and one output layer. A total of 10 models were trained (5 LSTM models and 5 GRU models), with the optimal timestep determined for each model individually. The initial experiments indicated that both LSTM and GRU models faced challenges in predicting certain air pollutant components at specific stations. Therefore, a second round of experiments involving re-training and re-testing was conducted. Based on the final inputs and evaluation, it was found that, for this particular case, the GRU model outperformed the LSTM model.

Item Type: Thesis (Undergraduate)
Uncontrolled Keywords: polusi udara, LSTM, GRU, CRISP-DM
Subjects: T Technology > T Technology (General)
Divisions: Fakultas Teknik dan Kejuruan > Jurusan Teknik Informatika > Program Studi Ilmu Komputer (S1)
Depositing User: NI PUTU NITA NATHALIA
Date Deposited: 15 Jan 2026 01:53
Last Modified: 15 Jan 2026 01:53
URI: http://repo.undiksha.ac.id/id/eprint/27528

Actions (login required)

View Item View Item