PREDICTIVE MODELLING OF CLEAN WATER SUPPLY IN RIAU PROVINCE: A DEEP LEARNING APPROACH

Abstract

The supply of clean water remains a critical issue in many regions, including Riau Province, where factors such as population growth and climate variability significantly affect its availability and distribution. This study aims to develop a time-series–based predictive model for clean water supply in Riau Province using deep learning approaches. Using historical data from 2019 to 2023, including variables such as the number of customers, water volume, economic value, and input costs, this research identifies temporal patterns to support proactive water resource management. The methodology consists of exploratory data analysis, data preprocessing, and model training using several architectures, namely Long Short-Term Memory (LSTM), Gated Recurrent Unit (GRU), Recurrent Neural Network (RNN), Convolutional Neural Network (CNN), and Feedforward Neural Network (FNN). Among these models, the LSTM achieved the best performance, with a Mean Absolute Error (MAE) of 1.25, a Mean Squared Error (MSE) of 2.56, and an R-squared (R²) of 0.92. After hyperparameter optimization, further improvements in predictive accuracy were obtained. Based on the optimized LSTM predictive model, the forecasted clean water volume for 2024 is 19,496.90 thousand m³, a slight decline from the previous year. The novelty of this study lies in the comprehensive comparison of multiple deep learning architectures for regional-scale clean water time-series forecasting and the optimized implementation of LSTM for operational prediction. In practical terms, the results can support local water authorities in improving planning, infrastructure development, and demand management strategies. However, this study is limited by the use of secondary data from a single province and a relatively short observation period, which may affect the model's generalizability. The proposed predictive framework can serve as a reference for future studies in sustainable water resource management.