A Comparative Study on Criminal Cases through Economic Indicators in Malaysia using Regression Modelling

Abstract

Criminal occurred when someone committed a crime and was charged under the justice system for violating the public law. Criminal analysis is commonly applied with economics in modelling of criminal cases since sluggish economy causes citizens to have no proper earning income and forces them to commit crime to get back to their regular life. This research study on the suitable regression method for prediction of the criminal case model between ordinary least squares regression, LASSO regression and Ridge regression. A total of 51 observations of economic indicators were collected from the Department of Statistics Malaysia and Bank Negara Malaysia as well as criminal cases were collected from the Federal Court of Malaysia. Based on the selection features from ordinary least squares regression and LASSO regression, the study found out consumer price index and KLCI were the key variables that have the impact on criminal cases. However, there were additional two variables: unemployment rate and participation rate that selected by LASSO regression do have a significant impact on criminal cases. In addition, the prediction models obtained were found out to have insensible results due to the intercept of the models showing a negative criminal case value. These insensible results were caused by the fluctuation of data that occurred during the pandemic of COVID-19 outbreak in Malaysia started February 2020 which had largely affected the economy and social behaviours. Lastly, the result found LASSO regression allowed to use 4 economic indicators and given a lower mean squared error with less complexity of the model, then allowed to give a better R-squared value as compared to other regression models.