Crime Pattern Analysis and Prediction Using Machine Learning

Abstract

It has been observed a massive rising curve in crime in the country. Only if crime could be decreased, some concrete steps should be taken by the Government and hence, we aim to brief about the crime patterns and predict using machine learning algorithm for the data obtained from the official Indian Government websites. The raw data obtained was converted to a suitable format using data mining techniques such as eliminating missing values, eliminating redundant data, data transformation, etc. This data was fed to algorithms like Linear Regression, Random Forest for performing predictions. Crime type predictions are performed, for four years, for each state as well as all the states of India using the data from year 2010-2020. These predictions are displayed using simple visualization charts. One important aspect that is used with these algorithms is that of identifying the trend changing year in order to increase the accuracy of the predictions.

Another aspect of the proposed analytics system includes the visualization of crime patterns over the past years till 2010 to 2020. These visualizations are performed with the help of demographic data available at the Census India website. Each state in the visualizations is distinguished using factors like literacy and area in order to demonstrate the crime per one lakh people. While performing all these predictions and visualizations, we came across significant observations regarding the algorithms used, or certain crime types and the data related to its which has also been incorporated in one of our modules.

The entire system is established through the project - “Indian Crime Analysis”. The project report provides three main services - predictions, visualizations and observations. Under the predictions service, Indian crime is considered. The visualizations service consists of the total crime against both the gender with respect to States in different years. The observations service provides a headline of all the notable statistics and information perceived during execution of the system. This project provides a user-friendly environment for all its users to take the entire benefit of studying the crime patterns across all the states of India for diverse purposes.