Abstract:- This paper examines the current techniques that are used to predict crime and criminality. Crime prediction is an attempt to identifying and reducing the future crime. Crime prediction uses previous data and after identifying data, predict the future crime with location, pattern and time. Crimes are a social nuisance and cost our society dearly in several ways. Presently, serial criminal cases quickly occur, so it is an demanding task to predict future crime accurately with better performance. Data mining techniques are very useful to solving Crime detection problem. So the aim of this paper to study various computational techniques used to predict future crime. A more number of research papers on this topic have already been published pastly. In this paper, we thoroughly survey each of them and summarized the outcomes.

Crime is “An act committed or omitted in violation of a law forbidding or commanding it and for which punishment is imposed upon conviction”.
We have categorised crimes into two types:
? Major crimes: Murders, rapes etc.
? Volume crimes: Burglary(illegal entry of a building with intent to commit a crime), robbery, theft, vehicle crime damage etc.

We Will Write a Custom Essay Specifically
For You For Only $13.90/page!

order now

Crime analysis is sensitive domain where efficient for prediction & classification to analyze the increasing numbers of crime data. The biggest challenge facing by many law enforcement is how to efficiently and accurately analyzing the increasing volumes of data. Crime is prime concern of the paper. Crime can take place at any time and at any region of the country. Crime is a problem which must be tackled. Large number of peoples is affected by crime. Crime forecasting and extracting useful information from large amount of crime data. Hence is very important. A survey is conducted so that crime forecasting can be improved by the use of efficient data collection and data mining strategies. Crime predictions can be made through both qualitative and quantitative methods. Qualitative approaches to forecasting crime as environmental scanning, scenario writing, are useful in identifying the future nature of criminal activity.
In Data mining ,Clustering and Classification are generally used methods of analyzing, detection, identification, prediction of crime.
Classification: Classification is the most commonly used data mining technique, where a model is constructing to predict class. It is used to classify each item in a set of data into one of predefined set of classes or groups. Classification method uses mathematical techniques such as decision trees, linear programming, neural network and statistics. Classification classifies data items into groups. Types of classification models are Classification by decision tree induction, Bayesian Classification, Neural Networks, Support Vector Machines (SVM), and Classification Based on Associations.
Clustering: Clustering is a data mining technique that makes useful cluster of objects that have similar characteristic. Classification techniques can also be used for distinct groups or classes of object but it becomes costly so clustering can be used as preprocessing technique for selection and classification. Types of clustering methods are Partitioning Methods, Hierarchical Agglomerative (divisive) method; Density based methods, Grid-based methods, and Model based methods.


2.1 Journey to crime:
Journey to crime supports the notion that crimes are likely to occur closer to an criminal’s home. Criminal acts follow a Distance-Decay function(DDF), such that the further away the regular activity space of an offender is , the less likely that the person will engage predatory criminal activity. It is worried with the “Distance Of Crime” and that offenders will in general move limited distances to commit their crime1.

2.2 Circle theory:
In the circle method, the distances between crimes are measured and the two most distant crimes are chosen. Then, a extreme circle is drawn so that both of the points are on the great circle. The midpoint of this extreme circle is then the assumed location of the criminal’s residence and the area bounded by the great circle is where the criminal operates. This model is computationally economical and easy to understand. Moreover, it is easy to use and requires very little training in order to master the technique. However, it has some drawback. for example, the area given by this method is often very large and other studies have shown that a smaller area success. Additionally, a few outliers can generate an even search area, thereby further slowing the police effort2.

2.3 Routine Activity Theory:
Routine activity theory is a sub-field of crime opportunity theory that centre of attention on situations of crimes. The general situation of an individual plays an important role in the definition of routine activity theory. The more one is showed to criminal behavior in their everyday lifestyle, the higher the likelihood that a person will perform criminal activity. According to Cohen and Felson, the union of three elements in time and space are required for a crime to occur: a likely offender, a suitable target and the absence of a capable guardian against crimes.



Support Vector Machines(SVM) were introduced by Boser, Guyon, vladimir Vapnik, in the year 1992. SVMs have become popular because of their success in handwritten digit recognition. Support vector machine forms the new generation of machine-learning methods used to discover perfect distinguishable between classes with in datasets. SVMs are now important and active field of all Machine Learning research and Pattern classification. SVMs were developed to solve the classification problem, but recently they have been extended to solve regression problems 5. Support Vector machine (SVM) is a nonlinear classifier and it has been performed well for time series prediction 12. SVM has been performed well in prediction of time series because they can model nonlinear relations in a stable and efficient way 14.Predicting the common disease like diabetes and pre-diabetes and Hot-spots prediction can be done in Support Vector Machine..Thus, SVM was applied in predicting of crime hot-spots 16 and predict the common diseases (diabetes and pre-diabetes) 17.


A multivariate time series is a flow of data points, determine typically at successive points in time spaced at uniform time intervals. A multivariate time series is to predict the future values based on history of variations in the data. Multivariate time series data are available in many realistic state such as gainful data and substantial data are usually multi-dimensional.
A multivariate time series clustering technique based on dynamic time wrapping (DTW) has been proposed to find similar crime trends. The Multivariate Time Series Clustering based on Dynamic Time Wrapping is cost effective for discovering similar crime trends and also in for predict them appropriately.


A neural network is a circuit composed of a very large number of processing components that are called Neuron. Each component works only on local information. Moreover, each component operates nonparallerly, thus there is no system clock. Artificial neural network was applied in predicting the geo-temporal variations of crime and disorder 30. ANN model is based on the prediction by smartly examining the trend from an already existing large historical set of data 26.

The Fuzzy time series models developed by Song and Chissom. A fuzzy time series modeling needs less computational efforts and has outstanding learning capabilities. The Fuzzy time series technique was applied in order to detect a crime pattern in some region. To reduce the computational overhead, Chen 14 simplified the process and proposed a simplified model that includes only simple arithmetic operations. The characteristic of the fuzzy time series modeling is used to improve the prediction efficiency. The extent of the Fuzzy time series is limited to inquest the suitability of fuzzy time series method to predict the crime and was applied for forecasting crime 22, and to provide practical computational techniques. The approach used in this technique are also worked properly even if some data are not available.