Pengembangan Aplikasi Android untuk Prediksi Water Quality Index (WQI) dalam Aquascape Menggunakan Gradient Boosting Regression dan Seleksi Fitur Improved Grey Relational Analysis (IGRA)

Abstract

Water is one of the most important elements in an aquascape environment, where small changes in water parameters can affect the Water Quality and environmental balance in an aquascape. There are several ways to check Water Quality, but traditional models such as laboratory testing or the use of IoT sensors often require a lot of money and time, making them less suitable for aquascape users. Therefore, this research was conducted to assist aquascape users in monitoring their Water Quality periodically using Improved Grey Relational Analysis (IGRA) feature selection and the Gradient Boosting Regression (GBR) algorithm XGBoost variant. This research was conducted to develop an Android application that can predict Water Quality Index (WQI) based on manual input from water parameters, namely pH, DO, temperature, TDS, Ammonia, KH, CO2, and light intensity with a dataset consisting of 13,293 lines of data. The XGBoost model built using a training and test data ratio of 80:20 was able to provide excellent results. This is reflected by the evaluation results of the R-squared (R2) metric of 0.9975, Root Mean Squared Error (RMSE) of 0.5440, and accuracy of 99.39%. This research shows that the model can be an aid for aquascape users in monitoring Water Quality more effectively, so as to maintain the balance of the aquascape ecosystem.