Analisis Optimasi Algoritma Backpropagation Momentum Dalam Memprediksi Jenis Tingkat Kejahatan Di Kecamatan Tambusai Utara

Abstract

Crime is conduct that disobeys the law and contravenes social norms in a way that society finds objectionable. There is no system that can forecast the kind and quantity of crimes that will happen in the future as a guide in the process of preventing crime, according to the North Tambusai Police. However, the public service process in the form of complaints has been done digitally. Backpropagation is an iterative method that works well even with complex and convoluted data. Backpropagation is an artificial neural network with several levels (multi-layer). Data processing is done on the types and numbers of crimes that took place in North Tambusai District between 2015 and 2020. The first step in the data processing procedure is to normalize the data and choose the network training parameters. Crime data and levels were used in the artificial neural network research, which used a 5-5-1 design. The test results reveal that the average prediction accuracy rate is 92.66%, with the greatest prediction accuracy rate being 99.6% and the lowest forecast accuracy rate being 90.01 percent. Theft had the highest weighting (Curat) of crimes the next year with 15 cases, while fraud, crime, and extortion/threats each had the lowest weighting (1 case). The prediction findings exhibit a sufficiently high level of accuracy to serve as a basis for evaluation.