Crime Time-Series Forecasting in Indonesia Using Neural Network and Linear Regression

Crime is a harmful and widespread social issue affecting individuals globally, including in Indonesia, requiring not only reactive but also predictive measures for effective prevention. Lack of continuously updated official crime information and predictions can lead to a variety of adverse outcomes. Without current data, law enforcement agencies may struggle to allocate resources effectively, potentially leaving some communities underserved and vulnerable. Public safety could be compromised as patterns of criminal activity shift without being detected in real-time, preventing the implementation of proactive measures. Furthermore, the absence of accurate predictions hampers the ability to forecast and prepare for possible crime waves or emerging threats. This deficiency can also undermine public trust in law enforcement's capacity to protect citizens and maintain order. This research employed data mining and machine learning methods to forecast crime trends in Indonesia. The approach depends on continuously refreshed and accurate data derived from public court rulings. Data from the Supreme Court of Indonesia was gathered, stored, and pre-processed, after which various machine learning models were utilized to produce crime time-series forecasting. These predictions aim to provide law enforcement and the society with actionable insights for targeted crime prevention, enhancing public safety and preparedness. This research demonstrates the potential of data mining and machine learning in accurately predicting crimes, supporting efforts to safeguard society.