PERAMALAN INDEKS KUALITAS UDARA DI KOTA SURABAYA MENGGUNAKAN METODE LSTM DAN CNN-LSTM

Adyatma, Bagas Mujaddid (2025) PERAMALAN INDEKS KUALITAS UDARA DI KOTA SURABAYA MENGGUNAKAN METODE LSTM DAN CNN-LSTM. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5003211048-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (8MB)

Abstract

Kualitas udara merupakan aspek penting yang berdampak langsung terhadap kesehatan manusia dan lingkungan. Kota Surabaya sebagai kota metropolitan mengalami peningkatan aktivitas industri dan transportasi yang berpotensi menurunkan kualitas udara. Oleh karena itu, diperlukan model peramalan yang akurat untuk memantau dan memprediksi Indeks Standar Pencemar Udara (ISPU). Penelitian ini bertujuan untuk meramalkan ISPU di Kota Surabaya secara univariat dengan menggunakan dua pendekatan deep learning, yaitu Long Short-Term Memory (LSTM) dan hybrid Convolutional Neural Network–Long Short-Term Memory (CNN-LSTM). Data yang digunakan merupakan data harian dari Stasiun Pemantauan Kualitas Udara (SPKU) milik Dinas Lingkungan Hidup Kota Surabaya, dengan fokus pada lima parameter utama, yaitu PM₁₀, SO₂, CO, O₃, dan NO₂. Proses pemodelan dilakukan secara terpisah untuk setiap parameter melalui tahapan imputasi Kalman Smoothing, normalisasi Min-Max, penentuan lag optimal, dan tuning hyperparameter. Hasil evaluasi menunjukkan bahwa model CNN-LSTM secara umum memiliki performa lebih baik dibandingkan LSTM dalam meramalkan seluruh ISPU berdasarkan evaluasi MAE, MAPE, RMSE dan IA. Temuan ini menunjukkan bahwa pendekatan CNN-LSTM dapat diandalkan sebagai dasar peramalan kualitas udara dan mendukung upaya pemantauan lingkungan secara berkelanjutan di Kota Surabaya.
===================================================================================
Air quality is a crucial aspect that directly affects human health and the environment. As a metropolitan city, Surabaya experiences increasing industrial and transportation activities, which potentially degrade its air quality. Therefore, an accurate forecasting model is needed to monitor and predict the Air Pollutant Standard Index (ISPU). This study aims to forecast ISPU in Surabaya using a univariate approach with two deep learning methods: Long Short-Term Memory (LSTM) and a hybrid Convolutional Neural Network–Long Short-Term Memory (CNN-LSTM). The data used are daily records from the Air Quality Monitoring Station (SPKU) operated by the Surabaya Environmental Agency, focusing on five main parameters: PM₁₀, SO₂, CO, O₃, and NO₂. The modeling process was conducted separately for each parameter through data preprocessing (Kalman Smoothing imputation and Min-Max normalization), determination of optimal lags, and hyperparameter tuning. The evaluation results indicate that CNN-LSTM generally performs better than LSTM in forecasting all ISPU values based on the MAE, MAPE, RMSE, and IA evaluations. These findings suggest that the CNN-LSTM approach is a reliable foundation for air quality forecasting and supports sustainable environmental monitoring efforts in Surabaya.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Polusi Udara, Kota Surabaya, Peramalan Kualitas Udara, CNN, LSTM, CNN-LSTM, Air Pollution, Surabaya, Air Quality Forecasting, CNN, LSTM, CNN-LSTM
Subjects:	H Social Sciences > HA Statistics > HA30.3 Time-series analysis
Divisions:	Faculty of Science and Data Analytics (SCIENTICS) > Statistics > 49201-(S1) Undergraduate Thesis
Depositing User:	Bagas Mujaddid Adyatma
Date Deposited:	04 Aug 2025 06:35
Last Modified:	04 Aug 2025 06:35
URI:	http://repository.its.ac.id/id/eprint/126262

Actions (login required)

View Item