PREDIKSI TINGKAT KEJAHATAN BERDASARKAN ARTIKEL BERITA NASIONAL MENGGUNAKAN METODE SUPPORT VECTOR MACHINE (SVM)

Abstract

Crime is any act or act that is economically and psychologically detrimental, violates the laws and regulations in force in the Indonesian province, as well as social and religious norms. Society opposes it because it can be interpreted as a crime that violates laws and social norms. Research This study aims to predict criminal acts which include narcotics, murder, theft, robbery and maltreatment. The data was taken from online media websites such as detik.com, okezone.com, tribunnews.com and kompas.com. The crime data was taken using a scraping technique. The method used in this research is the Support Vector Machiner algorithm by utilizing the python programming language with jupyter tools. From testing data testing as much as 30%, it shows a dataset with trend, seasonal and residual variables using the support vector machine algorithm with the Radial Basis Fauction kernel resulting in RMSE of 3,132 on murder crime data, 0.32 on prediction of robbery crimes, 2,017 on crime prediction cases. theft, 2.49 on the prediction of the crime of assault cases and 2.49 on the prediction of the crime of narcotics cases. Thus, from the results of model testing, it can be concluded that the trend, seasonal and residual variables are accurate and have the lowest RMSE value