Predicting Countries with Low and High Robbery Rates Using Discriminant Analysis

Abstract

Crime such as robbery has been identified as one of the socioeconomic problems across the world, which adverse social, economic, and family conditions have caused. Using discriminant analysis, this study proposed a model for classifying and predicting countries with low and high robbery rates. Robbery rates in 2018 of 42 countries across the world have been extracted from United Nations Office on Drugs and Crime as the dependent variable. Meanwhile, the independent variables included the unemployment rate, average household size, and poverty index. The study originally classified 32 countries with low and 10 with high robbery rates. Pretesting was employed, and the results showed that all the assumptions for discriminant analysis were fulfilled. Using standardized beta and Wilk's Lambda, the average household size is the best predictor variable, while the unemployment rate is the least predictor variable. The overall prediction function model is significant. The classification results by discriminant analysis algorithm for groups with low and high robbery rates show that the proposed model correctly predicts 78.6% of the robbery rates of countries based on the three characteristics, such as their unemployment rate, average household size, and poverty index.