A MULTI-DIMENSIONAL CRIME INDEX FOR CRIMES AGAINST WOMEN USING A HYBRID MACHINE LEARNING AND REGRESSION APPROACH

Abstract

The increasing concern over crime rates, particularly crimes against women, has led to a demand for more effective ways to assess and address public safety. Historically, crime analysis has relied on traditional statistical methods and crime reporting systems, which primarily depend on police reports, surveys, and judicial records. These systems, while valuable, suffer from limitations such as underreporting of crimes, biases in reporting, and a lack of integration across different sources of data. Moreover, traditional crime indices often focus on a single dimension of crime, overlooking the complexity and multi-dimensional nature of crimes against women, which include physical violence, sexual assault, harassment, and emotional abuse. These shortcomings make it difficult to create a comprehensive understanding of the safety issues faced by women. The problem arises in the need for a more holistic approach to crime analysis that integrates multiple dimensions of crime data, such as geographical location, socio-economic factors, time, and severity, while also considering the disparities in reporting and recording crimes. The absence of a multi-faceted crime index specifically designed to address crimes against women hampers effective policy-making and resource allocation for prevention and intervention programs. The motivation for this study stems from the growing need to develop more accurate and nuanced crime indices that reflect the complexities of crime against women. By incorporating diverse data sources and analyzing various factors influencing crime, a multi-dimensional crime index can provide valuable insights for law enforcement, policymakers, and social organizations. To address these issues, a proposed system combines hybrid machine learning and regression approaches to validate a multi-dimensional crime index. This system would integrate different data types, such as demographic, geographical, and crime-specific details, allowing for a more accurate assessment of the crime landscape. This comprehensive approach aims to improve both the reliability and the predictive capabilities of crime indices, offering a tool for more informed decision-making.