Use of Data Mining Technology to Identify Narcotics Distribution Patterns

Abstract

The abuse of narcotics has become one of the significant social and health problems in various countries worldwide. Conventional methods relying on manual analysis or traditional approaches may not be effective enough in addressing this challenge. Therefore, a more sophisticated and efficient approach is needed to tackle this issue. Data mining uses techniques from statistics, machine learning, and pattern recognition to extract valuable information from large data sets. This research employs data collection methods from the Narcotics Investigation Directorate of the Maluku Regional Police from 2021 to 2023. This data includes profiles of narcotics users, such as the age of the perpetrators, gender, last education level, occupation, location of arrest, and type of narcotic. The aim is to identify the patterns of narcotic distribution in the Maluku Province using data mining techniques, namely the Apriori algorithm, Naive Bayes, Random Forest, and Support Vector Machine (SVM). The exclusion of the age variable was a correct decision, as it resulted in an increase in accuracy. This increase is likely due to the high variation in the age variable. The accuracy improvement was more evident in the Random Forest algorithm compared to Naive Bayes and SVM. Random Forest achieved satisfactory results with an accuracy of 0.96. This indicates that Random Forest is a good algorithm for predicting narcotics user data. These results suggest that the pattern of narcotics distribution is closely associated with specific factors, including the male gender, the highest level of education being high school, a self-employed occupation, arrest locations on public roads, and the type of narcotic being Shabu.