NBKR INSTITUTE OF SCIENCE AND TECHNOLOGY

(Vidyanagar, Kota, Tirupati)

Department of Computer Science & Engineering

Prediction of Crime hotspots using machine learning

algorithms
BATCH – C10
Presented by:

V.Bhavana – 19KB1A05I0
V.Dharani – 19KB1A05I2
SK.Shakeer – 19KB1A05F4
R.Manoz – 19KB1A05D6
Under the esteemed Guidance of
Mr. V.Sai Charan , M.Tech
Assistant Professor, Department of CSE
CONTENTS
 Abstract

 Motivation

 Problem Statement

 Existing system

 Proposed System

 Methodologies

 UML Diagrams

 Data Flow Diagram

 Output Screen

 Conclusion

 Future Enhancements
ABSTRACT
Crime prediction is of great signification to the formulation of policing
strategies and the implementation of crime prevention and control. Machine learning is
the current main stream prediction method However Few studies have systematically
compare different machine learning methods for crime prediction .This project takes the
historical data of public property crime from 2015 to 2018 from a section of a large
coastal city in the southeast of China as research data to asses the predictive power
between several machine learning algorithms results based on the historical crime data
alone suggest that the Random Forest model outperformed KNN, support vector
machine , naïve Bayes and convolutional neural networks, in addition the built
environment data of points of interests(POIs) and urban road network density are input
into Random Forest model as covariates . it is found that the model with built
environment covariates has better prediction effect compared with the original model that
is based on historical crime data alone. The focus of crime hotspot prediction is to
forecast future concentration of criminal events in geographical space. Theoretical
criminology provides the necessary theoretical basis. Specifically ,several related
criminological theories not only provide guidance for us to understand the important
influence of location factors in the formation and aggregation of criminal events, but also
provide a basic mechanism for the police to use information of crime hotspots for crime
prevention or control .
MOTIVATION
 The main motive of this project is to predict the Crime Hotspots using various

Machine learning algorithms.This will help the people to take precautions before

going to such places.Also it helps the police to take necessary actions especially in

those areas with high crime rate. The focus of crime hotspot prediction is to forecast

future concentration of criminal events in geographical space

PROBLEM STATEMENT
Crime is one of the biggest and dominating problem in our society and its
prevention is an important task. Daily there are huge numbers of crimes committed
frequently.This requires keeping track of all the crimes and maintaining a database
for same which may be used for future reference. The current problem faced are
maintaining of proper dataset of crime and analyzing this data to help in predicting
and solving crimes in future.

Crime hotspot prediction aims to predict the likely location of future

crime events and hotspots where the future events would concentrate.

Nowadays people are using different algorithms to obtain the data of

crime hotspots but they don’t know which one is the most effective and accurate.
EXISTING SYSTEM
Many studies have carried out empirical research on crime prediction in different
time periods, combining demographic and economic statistics data, land use data, mobile
phone data and crime history data. Crime hotspot prediction aims to predict the likely
location of the future crime events and hotspots where the future events would
concentrate. A commonly used method is kernel destiny estimation. A model that considers
temporal or spatial autocorrelations of past events performs better than those that fail to
account for the autocorrelation. Recently Machine learning algorithms have gained
popularity.
Disadvantages
 lack of ability to defend in time and space.
Less accuracy
PROPOSED SYSTEM

In this project, random forest algorithm, KNN algorithm, SVM algorithm and
Decision tree algorithm are used for crime prediction.First, historical crime data alone are
used as input to calibrate the models. Comparison would identify the most effective
model.Second, built environment data such as road network density and poi are added to
the predictive model as covariates, to see if prediction accuracy can be further improved.
Advantages
1. The prediction results are better than those of the original model using historical crime
data alone.
2. Our models have improved prediction accuracies, compared with other models.
SYSTEM ARCHITECTURE
METHODOLOGY
 The steps that are commonly involved are:

1. Upload Crime dataset

2. Data Preprocessing

3. Feature selection

4. Building and Training model

5. Prediction

6. Visualization
1. Upload Crime Dataset:
Using this module we will upload crime dataset to application and then plot graph
of total crimes on different locations and then read dataset
2. Preprocess Dataset:
Using this module we will process dataset to convert all non-numeric data to
numeric data.
3. Run SVM Algorithm:
Using this module we will trained SVM algorithm on processed dataset and then
predict on test data and then calculate HIT RATE between original data and predicted
data.
4. Run KNN Algorithm:
Using this module we will trained KNN algorithm on processed dataset and then
predict on test data and then calculate HIT RATE between original data and predicted
data.
5. Run Random Forest Algorithm:

Using this module we will trained Random Forest algorithm on processed

dataset and then predict on test data and then calculate HIT RATE between original
data and predicted data.

6. Run Naïve Bayes Algorithm:

Using this module we will trained Naïve Bayes algorithm on processed

dataset and then predict on test data and then calculate HIT RATE between original
data and predicted data.

7. Run Decision Tree Classifier Algorithm:

Using this module we will trained Decision tree algorithm on processed

dataset and then predict on test data and then calculate HIT RATE between original
data and predicted data.
8.Run Logistic Regression Algorithm:

Using this module we will trained Logistic Regression algorithm on

processed dataset and then predict on test data and then calculate HIT RATE
between original data and predicted data.

9. Hit Rate Comparison Algorithm:

Using this module we will plot hit rate comparison graph between all
algorithms
UML DIAGRAMS
Use Case Diagram
Sequence Diagram
Activity Diagram
Class Diagram
Data Flow Diagram
OUTPUT
CONCLUSION
In this project, six machine learning algorithms are applied to predict
the occurrence of crime hotspots in a town in the southeast coastal city of China.
The following conclusions are drawn:
 The prediction accuracies of Random Forest model are better than those of the

other models. It can better extract the pattern and regularity from historical
crime data.
 The addition of urban built environment covariates further improves the

prediction accuracies of the Random Forest model. The prediction results are
better than those of the original model using historical crime data alone. Our
models have improved prediction accuracies, compared with other models.
FUTURE ENHANCEMENTS
For the future research, there are still some aspects to be improved. The first is the
temporal resolution of the prediction. Felson et al. revealed that the crime level changes with
time some studies have shown that it is useful to check the variation of risks during the day.
We chose two weeks as the prediction window. It does not capture the impact of crime
changes within a week, let alone the change within a day. There is no viable solution to this
challenging problem at this time. The second is the spatial resolution of the grid. In this
project, the grid size is 150m 150m. Future research will assess the impact of changing grid
sizes on prediction accuracy. Third, the robustness and generality of the findings of this
project needs to be tested in other study areas. Nonetheless, the findings of this research
have proven to be useful in a recent hotspot crime prevention experiment by the local police
department at the study size.
THANK YOU