MACHINE LEARNING: PROSPERITY OF RAINFALL, WATER DISCHARGE, AND FLOOD WITH WEB APPLICATION IN DELI SERDANG

Abstract

Flood event predictions can provide information to the surrounding community to prepare themselves for future. With the development of informatics, currently web-based applications are very accessible. PHP (Hypertext Preprocessor) is a programming language in the form of a script that can be implemented dynamically with HTML. PHP is used to build web-based applications and is implemented with other software. Software R is a command line based application that can be used to complete Machine Learning calculations quickly. In this study Backpropagation Neural Network (BP-NN) is used to predict rainfall and water discharge. Whereas Support Vector Machine (SVM) is used to predict flood events. The case study data used was Deli Serdang District in North Sumatra which often flooded. In this study, rainfall data was taken from three points or stations. The nearest river water debit is used to also affect flood events. Ensemble Machine Learning (BP-NN and SVM) uses the PHP programming language and R software is used for prediction. Using rainfall data from Kualanamu station, Tuntungan and Sampali as well as Sungai Ular water debit 1 January 2016-31 December 2017, the accuracy of flood prediction from this application is 94.4%.