Urooj Shehzadi (um8735091@gmail.com)
College of Statistical and Actuarial Sciences, University of the Punjab, Lahore, Pakistan
Maryam Ilyas (maryam.stat@pu.edu.pk)
College of Statistical and Actuarial Sciences, University of the Punjab, Lahore, Pakistan
Aiman Tahir (aimantahir78@gmail.com)
College of Statistical and Actuarial Sciences, University of the Punjab, Lahore, Pakistan
Abstract
This study is designed to interpret the yearly crime data of Pakistan at the country level for the period January 1997 to December 2018 and at the provincial level for the period January 1998 to December 2018. In crime analysis, the crime variables are typically large in number and are linearly related to each other. It is often difficult to interpret this type of dataset. To cope with these two problems, the Principal Factor Analysis (PFA) and correlation analysis are usually performed. Correlation analysis observe the pattern of relationship between the set of variables and PFA is a technique for reducing the dimensionality of such datasets, increasing interpretability but at the same time minimizing information loss. This study constructs the new transformed crime factors of original crime incidents that are low in dimensions and provide most of the total variability of the original variables. Crime against person and crime against property are two principal factors which are identified in this case study. These two factors explain 84.066% of the total variability in the data analysis of Pakistan. The 79.819% of the total variance of original variables is captured by these two factors in the crime analysis of Punjab.
Keywords: crime analysis, principal component analysis, factor analysis, dimension reduction
