Crime Pattern Detection Using Historical Data

Computer Science Engineering

The objective of the project is to create a platform where the police or responsible authorities can predict the location and observe patterns of crime. With help of crime data modeling the crime pattern can be detected or worked upon. Crime data analysis can even speed up the resolution to the court cases with ease and timely. Such types of crime prediction approach work on patterns of crimes in terms of similarity of attack or crime such as time of incidence, the methodology of crime is same etc. Collect the data: In this phase the data from various sources is collected from various government sources, social media platforms about the incident like Facebook and blogs. It may expect that the data received may be in unstructured form with different types and sizes. Classification; The standard algorithm useful for classification which may be applied is Naïve bayes Classifier. This classifier will determine the probability of falling into different classes of crimes and the crime predicted to belong to a specific class the probability is highest. Identify pattern: Next phase in the methodology is to find the sequence of crimes which are similar in nature and belong to the same class. Such a pattern may be identified as suggested in literature through the apriori algorithm. For the prediction outcome the decision trees are used. Here in the decision tree each internal node has a test for occurrence on an incidence and outcome are normally yes or no and based on the outcome again the next level of internal node has another question test and so on to reach a final decision for prediction. This type of analysis on historical data can also help in determining the criminal profile based on the characteristic behavior inferred from the data. It can help the investigator to accurately predict the profile of unknown criminals.

This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.