Authors:
- Katarina Tolja – University of Rijeka Faculty of Engineering, Department of Automation and Electronics, Rijeka, Croatia
- Jelena Musulin – University of Rijeka Faculty of Engineering, Department of Automation and Electronics, Rijeka, Croatia
- Daniel Štifanić – University of Rijeka Faculty of Engineering, Department of Automation and Electronics, Rijeka, Croatia
- Zlatan Car – University of Rijeka Faculty of Engineering, Department of Automation and Electronics, Rijeka, Croatia
Article type:
Original Scientific Paper
Abstract:
Drug use disorder is one of the leading health problems in the world, which as a medical and social phenomenon has been attracting general attention for many years. Personal motives for which an individual decides to consume drugs vary. In this research, the dataset for drug users classification consists of information from 1885 respondents and their usage of 18 drugs, both legal and illegal. Due to data imbalance, the research is based on three substances: ecstasy, cannabis, and nicotine. Three artificial intelligence algorithms—k-Nearest Neighbors (KNN), Support Vector Machine (SVM), and Multi-layer Perceptron (MLP)—were used to solve the classification problem. As a result, MLP achieved the highest AUC value compared to SVM and KNN.
Keywords:
binary classification, k-Nearest Neighbors, Support Vector Machine, Multi-layer Perceptron