Survey on Crime Interpretation and Forecasting Using Machine Learning

International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056
Volume: 09 Issue: 04 | Apr 2022 www.irjet.net p-ISSN: 2395-0072
© 2022, IRJET | Impact Factor value: 7.529 | ISO 9001:2008 Certified Journal | Page 1344
Survey on Crime Interpretation and Forecasting Using Machine
Learning
Ankita A Khartmol1, Nikki2, Swastika Arya3, Trushika Arya4
Mahalakshmi Manasa5
1 Student, Dept. of Computer Science engineering, Dayananda Sagar College of Engineering, Karnataka, India
5Professor, Dept. of Computer Science engineering, Dayananda Sagar College of Engineering, Karnataka, India
---------------------------------------------------------------------***---------------------------------------------------------------------
Abstract - Understanding the crime pattern and taking
safety measures to solve the crime problems has become the
important factor in today’s world. The main aim here is to
locate the crime location and find the pattern based on the
time and place in Bangalore city. There is a lot of rise in the
software systems which helps the officers to solve the crime
issues. We will be looking at the various machine learning
algorithms which helps to detect the crime pattern and helps
in solving the crime in very less time. There are different
algorithms in machine learning like k-means clustering
algorithm which can be used in the process of identification of
crime. The Data mining approach is used to predict the
features of crime dataset that affects the high crime rate.
There are two types of algorithms i.e., supervised and
unsupervised, supervised algorithms that is used to evaluate
the training dataset for its accuracy and unsupervised
algorithms helps in solving the unlabelled data into classes or
clusters. There are other few different learning methods like
random forest in data mining based on unknown and the
previous year collected datasets which can be used for
predicting the crime pattern based on the time and place.
Key Words: Crime data, Decision tree, KNN, Machine
learning, Naïve bayes and Random Forest
1. INTRODUCTION
Criminal activities are one of the most important and
common problems in the existing societyandtheprevention
of it is a key hassle. A huge number of crimes take place each
day. With the present scenario,nosolutioncangrantthebest
safety to an individual to prevent any harm that can be
caused by such an activity. Therefore, this necessitates
prediction of crimes in advance by storing all such criminal
data and maintaining a good database for the future and for
further analysis. Crime has become the huge problem in our
society and its safety is an important task. MachineLearning
algorithms have been increasing that made crimeprediction
possible based on the past data. To improve more security
measures in society and in explicit crime category regions
recognizing crime patterns will allow us to solve problems
with different approaches. Tounderstandthefuturescopeof
crimes past crime data trendingfactorsmayhelpustodetect
the future crimes. Gradually crime rate is increasing very
much. These days crimes are difficult to predict as they are
not systematic. Even the modern technologies and other
high-level techniques helps criminals in accomplishing the
crime. As per the crime records bureau crimes like robbery,
stealing, theft and other few have reduced but the other
crimes except these types of crime have been increased. It is
difficult to predict who may be the victims but it is possible
to predict the probability of occurrence. The outcome tells
that our application helps in reducing the crimetoa possible
extent by providing security in sensitive areas as the results
predicted cannot give clarity with full perfection. For
collecting the crime records and evaluating it, we need to
build a strong crime analytics tool. We will be using the
various ML models to predict crimes. To understand the
pattern of crimes we need to keep area wise track for the
geographical analysis. The other visualization techniques
and different plots can help law enforcementagenciesandto
the police department fields to predict the crimesanddetect
them with higher accuracy. By using data miningalgorithms,
it will be helpful to extract the previous unknown data and
useful information from the dataset and find out the crime
pattern. The challenges faced will be to analyse the dataset
and for maintaining a proper data of the criminal activities
and using the different statistics methodtosolveandpredict
the crimes in future. There is an approach to solve between
the criminal justice and computer science fordeveloping the
data mining procedure which helps to solvecrimefaster. We
will be focusing on the crime factors on each day instead of
finding out the criminal background of the offender and
other politics etc. The aim here is to analyse the data that
integrate a vast number of crimesandtoidentifythenumber
of crimes which may occur in future. Therefore, the crime
evaluation a prediction are well known approaches for
finding out and analysing the crime styles and the trending
factors in crime.
Figure – 1: Model Implementation

2. MOTIVATION
With high growth in popularity of crime analysis and
forensics, engineers and researchers havebeenabletofinda
large number of patterns taking place in the crime. The
meaning of a crime is any act which is punishable by law
under the government, since it is a gross violation of thelaw.
It refers to several variants of illegal activities that are
forbidden by law. Most crimes are known to be committed
today by an individual for personal pleasure or satisfaction,
lack of monetary benefits, lack of basic necessities such as
food and shelter, or for seeking revenge. Machine learning
now provides us with the mechanism to use artificial
intelligence to create automated analysis techniques that
were considered impossible for computers. The danger of
crimes occurring is high for a lot, especially in developing
countries where millions face poverty, dwell in slums,
unemployment and are hence exposed to increased risky
environments. Criminal actions are defined by the laws of
specific jurisdictions and many times, there are vast
differences between countries and even within the same
country regarding what kind of activity is prohibited and
illegal and can be classified as criminal. Recently, the
National Criminal Records Bureau of India accounted that
2,82,171 criminal cases were reported in a single year in
Uttar Pradesh alone, the highest for any state in India.
Therefore, it is important to classify what activities
constitute criminal offences and hence, the preventionof the
same and to safeguard the public is thus need of the hour.
3. RELATED WORK
[A] Lalitha and Suresh Babu, “Cluster based Zoning of Crime
Info”, (2017)
A CICD (Criminal Identification and Crime Detection)
method is used to identify illegal activities in India. In this
method, criminals are identified based on their facial
recognition, type of crime and its motive, weaponusedetc.It
consists of six primary steps: 1) Data extraction 2)
Clustering, 3) Pre-Processing, 4) Mapping, 5) Classification
and the use of WEKA tool (Collection of ML Algorithms). In
this method, K-Means and KNN together are used for
filtration in big data collections. In the United Kingdom,
Cambridge PoliceDepartmenthaveperformeda project with
the name Series Finder to find out the crime patterns in
theft. Patterns in crime have been extracted to attain the
criminals by using modulus operandi. The information
included are the way the criminal illegally enters the area of
the theft, timestamp of theft and the kind of property
(independent, lease, rental), along with the geographical
location with the other break-ins. Using different crime
series, it was able to identify most kind of crimes happening.
The prediction’s accuracy was high (above 80%) in London,
United Kingdom, mobile activities along with the location
information is speed to predict important crime spots. In
USA, Machine learning is applied on live stream of video
coming from CCTV footage to check any unusual activity
occurring, for example, K means algorithm is used in a room
which allows limited access to check if any unusual access
has occurred at a particular time. KDD – a collective method
which includes statistical modelling using machine learning
algorithms, one of the most important techniques in any
machine learning application, is used in crime analysis. The
main function of KDD is prediction of human behaviour and
finding patterns. [1]
[B] Chris Delaney, “Crime Pattern Definitions for Tactical
Analysis”, (August 2011)
Criminal patterns are a foundational concept in crime
analysis. By elaborating and classifying the various crimes
occurring in patterns, the IACA looks to increase
communication and enhance knowledge amongst police
practitioners. [2]
[C] Prajakta Yerpude and Vaishnavi Gudur, “Predictive
Modelling of Crime Dataset Using Data Mining”, (July, 2017)
Due to increase in crimes across the world, there is a need to
decrease the occurrence of such events that threaten
the right to one’s life. It helps society and cops to take
adequate steps to solve crimes faster. Data mining approach
has been taken to predict the crime dataset that result in
high crime rates. Regression, Naive BayesandDecisiontrees
are machine learning methods performed on past data and
used for forecasting that cause crime in a locality. The Police
Department and CrimesRecordBureaucantakemeasuresto
reduce crimes rates using the crime patterns and results
generated through predictions. [3]
[D] Shyam Nath, “Crime Pattern Detection Using Data
Mining”, (2006)
Data mining is used as a pattern for crime inspection
problems. Any study thatcan assistinresolvingcrimesfaster
will remedy the cruel acts in society. 10% of the criminals
constitute over 50% of the crimes that occur. Here,
clustering algorithm approachtoidentifythecriminal trends
is used to fasten up the process of resolving offences. Then
further K-means clustering has been taken and enhanced it
to predict crime. [4]
[E] Dawei Wang, Wei Ding, Henry Lo, Tomasz, “Crime
hotspot mapping using the crime related factors”, (2012)
The concept of the paper helps in analysing spatial factorsof
activities revolving crime. Spatial distribution of crime
involves a distribution of crime opportunity and socio-
economic factors. Current approaches live more on the
density of crime to store the information without analysing
the factors. A new crime hotspot mapping tool - Hotspot
Optimization Tool (HOT) is introduced. Itisanapplicationof
spatial data mining. HOT is Geospatial Discriminative

Patterns (GD Patterns) that captures dissimilarity between
the datasets. [5]
[F] Shiju Sathyadevan, Surya S, “Crime analysis and
prediction using data mining”, (2014)
The main objective is to focus on crime factors instead of
crime occurrences. With the help of data mining, previous
unknown data is extracted from raw data. One can
synthesize between criminal justiceandcomputerscience to
model a data mining technique to resolve crimes faster.This
paper explains various types of crime prediction and
analysis using several data mining. [6]
[G] Sathyadevan, S., & Gangadharan, S., “Crime Analysis and
Prediction Using Data Mining”, (2014, August)
Paper let us know about the cluster analysis which is done
by k-means algorithm on crime dataset. The custom map
generates hotspots of crime and displays crime details of
different states which helps the crime departments to take
precautionary measures to fight against the crime plan
advanced investigation strategies. Zoning awareness could
aid in cautioning police to take increased levels of action. [7]
[H] Tayebi, M. A., Gla, U., & Brantingham, P.L., “Learning
where to inspect: Location Learning for Crime Prediction”,
(2015, May)
The motive here is to look into the machine learning
methods for predicting crime which helps to yield better
results with good accuracy using better techniques with
importance of the crime datasets that helps in analyzing the
crime. Data set is analyzed with the help of the supervised
machine learning techniques to collect the information with
labelled data and do further validation process of it and
visualization will be on the complete dataset. [8]
[I] P. Thongtae and S. Srisuk, “An analysis of data mining
applications in crime domain”, (2008)
Examining and evaluating pre-existing databases and
evaluating them to generate meaningful information
underlies data mining. New information involving crime
trends is extracted and analysed from criminal dataset.
Increased number of techniquestopredictandtodoanalysis
in data mining is taken place. However, only a rare few
efforts were invested in the field of criminology. [9]
[J] Yadav, S., Timbadia, A., Vishwakarma, R., & Yadav, N.,
“Crime Patteren Detection, Analysis & Prediction”. (2017)
Series finder, a technique to find patternsincriminal activity
was developed by the police department, Cambridge UK.
They used the modus (M.O.) of offender as well as extracted
illegal patterns that were seen all the way through the
person responsible for. The police built M.O. of the guilty. It
is a fixed behaviour of a criminal and is variant of action
which represents a typeof pattern.Theinformationincluded
is the way the criminal illegally enters the area of theft, the
timestamp of theft and kind of property (independent,lease,
rental), along with the location etc. It identifiesseveral ofthe
majority of crime patterns and styles and also identifies
several illegal activities. The correctness of it was higher
than 80%. Therefore, a similar kind of approach is to be
implemented by finding patterns in crime. [10]
4. COMPARISON TABLE
Author and
Year
Title Concept Limitations
Shiju
Sathyadevan
,
Surya S,
2014
Crime
Analysis
and
Prediction
using Data
Mining:
Crime
analysis has
been a
strong and a
valuable
technique
for analysing
crime and
other
factors. Here
we can
extract the
data from
the unknow
sources.
To get better
results for
prediction there is
a need to make a
search for many
crime attributes
that is of different
places not by just
fixing it. Need to
include more
attributes for
better accuracy.
Prajakta
Yerpude
and
Vaishnavi
Gudur,
2017
Predictive
Modelling
of Crime
dataset
using data
mining
As there is
continuous
increase in
crime across
the globe,
crime data
should be
analysed to
decrease the
crime rate.
Therefore, it
helps the
police force
in solving
the crime
issues faster.
Higher the valueof
accuracy the data
become
insignificant for
the model
compared to
significant data
which shows that
regression needs
continuous data
including the
sparse values.
P. Thongtae
and S.
Srisuk,
2008
An
analysis of
Data
Mining
Applicatio
ns in
Crime
Domain
This paper
gives the
efficient and
effective
way to solve
the crime
problems
i.e., the
methods or
and different
types of
approaches
on data
mining for
crime data
analysis.
Here the
idea is to
find the
The data mining
techniquesandthe
meta- data has to
be chosen based
on background
awareness gained
of the analyst.
Hence, there is a
probability to
choose the
integrated data
which can be
performed
repeatedly.

criminal
activities of
professional
identity of
the fraud
criminals
based on the
cognizance
found of
their own
background.
Shyam
Varan Nath,
2006
Crime
Pattern
Detection
Using Data
Mining
In today’s
world
crimes have
been a social
bother
which cost’s
our society
in many
ways.
Researches
which can
help in
solving
crime faster
will be
effective.
Machine
learning
works with
geo-spatial
plot which
helps in the
improvemen
t in the
productivity
of the
detectives
and the
other crime
departments
.
Crime pattern
analysis can only
help the detective.
The quality of
input data is very
sensitive for data
mining which may
not be the exact
value and thatmay
have not had the
important
information.
Dawei Wang
Wei Ding,
Henry Lo,
Tomasz
2012
Crime
hotspot
mapping
using the
crime
related
factors—
Spatial
Data
Mining
Approach
The
technique of
hotspot
mapping has
been used in
vast for
analysing
the spatial
behaviour of
crimes. The
crime is
considered
to be the
related with
the socio-
economic
and other
community
of crime
factors.
The hotspot
mapping
technique is based
on the grid the
matic mapping
which doesn’t
permit for the
demonstration of
the hotspots. The
conversion of
points which
indicatesthecrime
incident cells with
crime counts and
the other details
across the cellscan
be lost.
Table – 1: Comparison Table
5. CONCLUSION
We can see just how easy it is to get started witha revolution
to safeguard our society from the evil eye, through crime
prediction which can change the whole crimescenariointhe
world. It is remarkable to see the success of machine
learning in such varied real-world problems. The aim of
society must not be just to catch criminals but to prevent
crimes from occurring in the first place. In the future, the
system could be extended to predict who will commit a
crime based on the behavioural patterns observed in the
person. It can also be used to correlate the relationship
between crimes occurring in different places and thereby
truly acting as a prevention to the public.Inthisproject,here
it has been demonstrated as a classification of the different
criminal acts wherein the user inputs a location and given
time for which he wants to predict if he is safe or not. The
model he, which separates it from other methods that rely
heavily on transfer learning approach and is user-friendly.
REFERENCES
[1] Lalitha and Suresh Babu, Cluster based Zoning of Crime
Info, IEEE, 2017, DOI:
ieeexplore.ieee.org/document/7905269
[2] Chris Delaney, “Crime Pattern Definitions for Tactical
Analysis”, (August 2011)
[3] Prajakta Yerpude and Vaishnavi Gudur, -- International
Journal of Data Mining & Knowledge Management
Process, Predictive Modelling of Crime Dataset Using
Data Mining, (IJDKP), Vol. 7, No.4, July 2017, DOI:
10.5121/ijdkp.2017.7404
[4] Shyam Nath, ―Crime Pattern Detection Using Data
Mining, IEEE/WIC/ACM International Conference on
Web Intelligence, 2006, DOI:
ieeexplore.ieee.org/document/4053200
[5] Dawei Wang, Wei Ding, Henry Lo, Tomasz, a spatial data
mining approach, Applied Intelligence, 2012, Josue
Salazar, Melissa Morabito, ―Crime hotspot mapping
using the crime related factors, DOI:
dl.acm.org/doi/10.1007/s10489-012-0400-x
[6] Shiju Sathyadevan, Surya S, ―Crime analysis and
prediction using data mining, First International
Conference on Networks & Soft Computing
(ICNSC2014), 2014, DOI: 10.1109/cnsc.2014.754321
[7] Sathyadevan, S., & Gangadharan, S. (2014, August).
Crime analysis and prediction using data mining. In
Networks & Soft Computing (ICNSC), 2014 First
International Conference on (pp. 406- 412). IEEE, DOI:
10.1109/CNSC.2014.6906719

[8] Tayebi, M. A., Gla, U., & Brantingham, P. L. (2015, May).
Learning where to inspect: location learning for crime
prediction. In Intelligence andSecurityInformatics(ISI),
2015 IEEE International Conference on (pp. 25-30).
IEEE, DOI: academia.edu/37571766/IRJET
[9] P. Thongtae and S. Srisuk, “An analysis of data mining
applications in crime domain”, IEEE 8th International
Conference on Computer and IT Workshops, 2008, DOI:
10.1109/CIT.2008.Workshops.80
[10] Yadav, S., Timbadia, A., Vishwakarma, R., & Yadav, N.
Crime pattern detection, analysis & prediction. In
Electronics, Communication and AerospaceTechnology
(ICECA),2017 International conference of (Vol. 1, pp.
225-230). IEEE, DOI: 10.1109/iceca.2017.8203676
[11] A. Bogomolov, B. Lepri, J. Staiano, N. Oliver, F. Pianesi
and A. Pentland, 'Once Upon a Crime: Towards Crime
Prediction from Demographics and Mobile Data', CoRR,
vol. 14092983,2014,DOI:10.1109/iceca.2017.8203575

Survey on Crime Interpretation and Forecasting Using Machine Learning

More Related Content

Similar to Survey on Crime Interpretation and Forecasting Using Machine Learning (20)

More from IRJET Journal (20)

Recently uploaded (20)

Survey on Crime Interpretation and Forecasting Using Machine Learning