An Emperical Investigation of Socio-Economic Parameters Associated with Crime in Karnataka Using Predictive Analysis

Abstract

The crime rate in India has been on a rise with growing population and rapid development. Crimes are a social nuisance and brings disrepute to the society and nation at large. Data mining techniques have enabled models to predict crime. The law enforcement officers and police personnel’s need to spend umpteen time to analyze crime from the crime reports published by National Crime Records Bureau and respective State police departments. Therefore, there is necessity of a mechanism which can predict crime by identifying factors responsible for increase in crime rate. This project is an attempt to address this. Socio-economic variables like poverty, urbanization, literacy, and the demographic and social composition of the population are recorded from Census data. Machine Learning algorithms namely Multiple Linear Regression, Random Forest Regressor and Generalized Linear Models are trained and deployed. These algorithms have been implemented assuming both linear and non-linear ship of data and subsequent results have been compared. Further, choropleth maps have been plotted to represent and understand data in a simple way. The results show that socio-economic factors are good indicator in explaining crime rates and good performance is observed with few models