Cloud Computing-based Shrimp Pond Water Quality Prediction Intelligent Service System

Maintaining water quality is an essential factor in the success of shrimp farming, particularly in conventional and semi-intensive methods in Indonesian. Poor water quality will affect shrimp's survival, reproduction, development, and harvest yield. In order to furnish data regarding future water quality conditions, This research aims to create an intelligent cloud-based water quality prediction system for shrimp ponds that can provide accurate predictions regarding future water quality conditions. The system utilizes the WQI dataset gathered from four different shrimp farming sites, totaling 408 samples, each location exhibiting a different set of values. The model will be trained using four parameters: pH, DO, salinity, and temperature. The WQI dataset will be pre-processed to address missing data, outliers, and standardization. The water quality prediction model uses three machine learning algorithms: SVM, ANN, and MLR. The model's performance results are evaluated using MAE, RMSE, and R². The results indicate that the ANN model is the most effective, achieving an MAE: 0.4023, RMSE: 0.5336, and R²: 0.7178 for temperature predictions, and an MAE: 0.4080, RMSE: 0.5942, and R²: 0.5997 for salinity. The SVM model had mixed results for temperature, with an MAE: 0.3645 and RMSE: 0.4823, but it performed poorly for DO, as evidenced by a negative R² of -0.2428. The MLR model provided reasonable temperature predictions MAE: 0.4953, RMSE: 0.6370, R²: 0.5602. Subsequent research endeavors should prioritize the augmentation of the dataset size and the incorporation of temporal dimensions in order to enhance the precision of predictive outcomes.