Real-time Water Level Monitoring System in Oriental Mindoro Using Neural Networks

Abstract

Oriental Mindoro faced an environmental problem that was caused by the massive rains from the highlands of the city. The lowlands cannot prepare for the upcoming floods with manual prediction and based on instinct. A water monitoring level was setup in three different rivers that end the streams in the city of Oriental Mindoro. This study focuses on the development of a real-time water level monitoring system to help with disaster preparedness in Oriental Mindoro. The dataset collected was from the provincial government, which shows the diversity of data that can be correlated to predict if a flash flood will occur. The dataset was tested using different algorithms such as Gradient Boosted Tree, SVM, Decision Tree, and Neural Networks. The prediction of the Neural Network model’s accuracy is 98.61%, which is 0.6 percentage points higher than the Gradient Boosted Tree and more than the rest of the algorithms.