PREDICTIVE ANALYSIS OF CRIMES AGAINST WOMEN IN INDIA

Abstract

In this thesis work, predictive analysis of crimes against women in India has been performed for crimes domestic violence, rape and dowry system. Time Series Forecasting is done using Gaussian Method, Linear Regression Model and Multilayer Perceptron Model using WEKA tool. ANOVA is used to compare the variances of the three methods to identify the most effective algorithm for prediction. Data has been collected from National Crime Records Bureau, which compiles all crime related data from all over country for research and analysis purposes. Various acts related to crime against women and the amendments done to them are also compiled. The analysis shows us how effective the laws are in curbing theses crimes. After assessing the results we found out that Multilayer Perceptron algorithm used for prediction is giving the most effective result as evident from ANOVA p- value. Crime rate, defined as the number of crimes per one lakh population, is used for analysis of crimes. ‘Crime rate’ for crimes committed against women for any year has been calculated using only female population based on projected mid-year female population for that year. Projected population is calculated based on population Census of India 2001, as supplied by Demographic Division, Registrar General of India, Ministry of Home Affairs. It has been observed that the number of dowry deaths in year 2013 is on a decline where as rape and dowry prohibition cases show an increasing trend after 2010.