PREDICTION OF LAGUNA LAKE WATER QUALITY IN TERNATE USING ARTIFICIAL NEURAL NETWORK

Abstract

Predicting the water quality of Laguna Lake in Ternate City is crucial to support ecosystem management and the needs of surrounding communities. This study aims to predict the parameters pH, temperature, nitrate, and phosphate for the year 2026 using four methods: Multiple Linear Regression (MLR), Random Forest Regression (RF), Support Vector Regression (SVR), and Artificial Neural Network (ANN). The data used consisted of measurements from 2017, 2023, and 2024. The prediction results show differences in performance among the methods. For pH, the 2026 predictions range from 6.54 to 8.15, with RF providing the highest value and ANN the lowest. Temperature predictions range from 31.03 to 31.86 °C, nitrate from 0.0888 to 0.1140 mg/L, and phosphate from 0.0433 to 0.0752 mg/L. ANN demonstrates the best accuracy for pH, while RF performs better for temperature. These differences indicate the presence of non-linear relationships among water quality parameters that are more complex than simple linear models. The 2026 water quality predictions offer an early assessment of Laguna Lake’s condition, providing a basis for informed management planning, conservation, and water quality decision-making. The results also highlight the effectiveness of artificial intelligence methods, particularly ANN, in modeling and predicting water quality conditions in tropical aquatic ecosystems.