Perbandingan Model LSTM Dan Random Forest dalam Peramalan Permintaan Air Di PDAM Surya Sembada Kota Surabaya

Inayah, Ghina Zulfa (2026) Perbandingan Model LSTM Dan Random Forest dalam Peramalan Permintaan Air Di PDAM Surya Sembada Kota Surabaya. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5003221090-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (4MB) | Request a copy

Abstract

Peramalan permintaan air merupakan aspek penting dalam mendukung perencanaan operasional dan pengelolaan sumber daya air pada perusahaan penyedia layanan air minum. Penelitian ini bertujuan untuk membandingkan kinerja metode Long Short-Term Memory (LSTM) dan Random Forest dalam meramalkan permintaan air pelanggan PDAM Surya Sembada Kota Surabaya serta memperoleh model terbaik untuk melakukan peramalan satu tahun ke depan. Data yang digunakan berupa data bulanan periode Januari 2010 hingga November 2025 yang mencakup permintaan air pelanggan, jumlah pelanggan, suhu minimum, suhu maksimum, tekanan udara, kelembapan, dan curah hujan. Sebelum pemodelan dilakukan, variabel eksogen diseleksi menggunakan uji kausalitas Granger dan karakteristik hubungan data dianalisis menggunakan uji Terasvirta. Pemodelan dilakukan dalam tiga skenario, yaitu multivariat, multivariat dengan fitur lag, dan univariat. Metode LSTM dibangun menggunakan teknik normalisasi, sliding window atau pembentukan fitur lag, serta hyperparameter tuning berbasis grid search, sedangkan Random Forest dikembangkan melalui sliding window atau pembentukan fitur lag dan optimasi hyperparameter menggunakan grid search. Kinerja model dievaluasi menggunakan Mean Absolute Percentage Error (MAPE) dan Root Mean Square Error (RMSE). Hasil penelitian menunjukkan bahwa model Random Forest Multivariat memberikan performa terbaik dengan nilai MAPE sebesar 2,888% dan RMSE sebesar 712.379,438 pada data uji. Hasil tersebut menunjukkan bahwa penggunaan variabel eksogen mampu meningkatkan akurasi prediksi permintaan air. Model terbaik selanjutnya digunakan untuk meramalkan permintaan air pelanggan PDAM Surya Sembada selama dua belas bulan ke depan. Hasil peramalan dapat digunakan sebagai dasar pendukung pengambilan keputusan operasional dan perencanaan kapasitas penyediaan air oleh pihak PDAM Surya Sembada Kota Surabaya. ================================================================================================================================
Water demand forecasting is an important aspect in supporting operational planning and water resource management in water supply companies. This study aims to compare the performance of Long Short-Term Memory (LSTM) and Random Forest methods in forecasting customer water demand at PDAM Surya Sembada in Surabaya City, as well as to obtain the best model for forecasting one year ahead. The data used are monthly data from January 2010 to November 2025, which include customer water demand, number of customers, minimum temperature, maximum temperature, air pressure, humidity, and rainfall. Prior to modeling, exogenous variables were selected using the Granger causality test, and the characteristics of the data relationships were analyzed using the Terasvirta test. Modeling was conducted under three scenarios: multivariate, multivariate with lag features, and univariate. The LSTM method was developed using normalization techniques, sliding window or lag feature construction, and hyperparameter tuning based on grid search, while the Random Forest model was developed using sliding window or lag feature construction and hyperparameter optimization using grid search. Model performance was evaluated using Mean Absolute Percentage Error (MAPE) and Root Mean Square Error (RMSE). The results show that the Multivariate Random Forest model achieved the best performance with a MAPE value of 2,888% and an RMSE of 712,379.438 on the testing data. These results indicate that the use of exogenous variables can improve the accuracy of demand prediction. The best model was then used to forecast customer water demand at PDAM Surya Sembada for the next twelve months. The forecasting results can serve as a supporting basis for operational decision-making and capacity planning of water supply by PDAM Surya Sembada in Surabaya City.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	LSTM, Random Forest, PDAM, Peramalan Permintaan Air, Time Series, LSTM, Random Forest, PDAM, Water Demand Forecasting, 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 > 49201-(S1) Undergraduate Thesis
Depositing User:	Ghina Zulfa Inayah
Date Deposited:	30 Jul 2026 01:08
Last Modified:	30 Jul 2026 01:08
URI:	http://repository.its.ac.id/id/eprint/138950

Actions (login required)

View Item