Mody University International Journal of Computing and Engineering Research

Volume 7 Issue 1, 2023, 31 - 34

Crime Rate Prediction using Enhanced K-Means

Algorithm
Rajeev Kumar
CSE Department, SET,
Mody University of Science and Technology, Lakshmangarh
raj83it@gmail.com

Abstract— Crime is incrementing every day. Crime is a truthfulness and association blunder parameters [12]. To
considerable constituent of the public. The growth of the recent raise the agglomeration method, from the subjective
technological influence, social media, and modern methods help alternatives, low-cost alternatives were deleted through
us to analyze crimes. There exist several kinds of clustering deciding on the proper threshold.
algorithms for crime prediction and analysis. Among these
clustering algorithms, I have proposed enhanced k-mean B. Crime Analysis using K-Means Clustering
algorithm for detection and analysis of crime data. It provides a
better method for analyzing and predicting the crime rate in Jyoti Agarwal et al. have conferred a paper on misdemeanor
particular areas. The most important application of this data forecast discrimination which shows k-approach, its
analysis is crime analysis. The main center of attention is on the
techniques and algorithms used for accurate prediction and compulsory oppression open supply information processing
analysis of data. After analyzing crimes, it helps us to predict tool which might be systematic gear used for reading
crimes. Instead of focusing on cause of the crime occurrence, we familiarity. Among the handy open supply facts processing
completely focus on the crime rate. suite like R, Tanagra, WEKA, KNIME, fast guide laborer
Keywords- Crime rate, Clustering algorithm, Data mining, - [13,14]. K shows that agglomeration is finished with the
means clustering. backing of instantaneous manual laborer tool this is
correlate open supply carried out math and facts
I. INTRODUCTION dispensation enclose written in Java with innovative
Crime can be defined as the act of doing illegal work which statistics mining guide picks.
is completely against principles like killing; raping, robbery
etc. can be considered as crime actions [1, 2]. In India the C. Parallel k-method set of rules on Distributed Memory
penalty for murder is lifetime custody [3]. For every crime multiprocessors
activity, there are severe set of laws and punishments will be
Joshi et al. have used the K-approach clustering records
useful according to the organization [4, 5]. Although state
and central governments are demanding to reduce the crime mining overall performance on the particular dataset, to
rate there is no matter what are the safety measures being categorize towns with a large contravention pace with the
taken by government the crime speed is mounting each aid of detecting contravention rates of each kind of brutality
distinct day of the week [6 , 7]. There will be lot of reasons [15]. The method used is dataset assortment supported
for the crime rate even though the government is trying to through preprocessing of statistics and followed by the
afford security procedures [8, 9]. The most common reasons scrutiny bearing on k-method using a clustering device
will be scarcity, drugs, political affairs, belief, family which includes, (a) acknowledgment concerning k, through
conditions, the public, unemployment, etc. Every year there
making use of silhouette appraise and (b) accumulation that
is an increase of 1.6% once a year enlarge of illegitimate
cases are being reported in India [10, 11]. Detection of statistics in the K-manner clustering tool. Later, the use of k-
crime rate in the approaching years can be dreadfully approach, cluster zero to cluster 4 are gained supported by
valuable for the government in categorizing the essential inspection concerning clusters received applying K-method
protection. in conjunction with the case observe almost about violation
in numerous regions.
II. LITERATURE REVIEW
D. Crime Analysis using K-Means Clustering
A. Analysis and Prediction of Crimes by clustering and
Nath used k-approach clustering to understand the styles in
Classification crime. Offenses comprehensively range in nature. Violation
RasoulKiani et al. Have conferred a substitution facts sets habitually enclose several unanswered crimes
construction for agglomeration and predicting crime [16,17]. The scenery of violations adjustments over
supported real recognition. Throughout this skeleton, the example. Fundamentally, the category system is based on
Genetic rule become habituated development outlier the current and known comprehended violations, and it will
recognition in the preprocessing department, and therefore now not supply exquisite perceptive eminence for destiny
the strength carry out turn out to be noted supported incorrect doings. Paying awareness to the beyond details,

©Mody University of Science and Technology, Lakshmangarh 31

Mody University International Journal of Computing and Engineering Research
Volume 7 Issue 1, 2023, 31 - 34

the author predicted that the clustering method is finer to changes clusters, otherwise homogeneously, until the
different supervised techniques like classification. centriods leftovers alike.
K-means clustering is a variety of cluster investigation
E. Analysis and Prediction of Crimes via Clustering and
which aims to separation of N observations into okay
Classification clusters wherein each remark belongs to the cluster with the
bordering suggest. The K-means clustering algorithm is
Rasoul K et al., analyzed the crime data the usage of the admired for the reason that it can be apply to reasonably
facts mining techniques together with K-way Algorithm for bulky sets of the statistics [23,24]. The client specifies the
grouping the similar crime patterns for identifying crime in
quantity of clusters to be originated. The algorithm then
modified years primarily based on amount of crime price at
separates the records into globular clusters by
some point of modified years and identifying the crime
pronouncement a set of cluster centers, conveying each
styles and dispositions to advise this manner can be used to
observation to a cluster, shaping new cluster centers and
cut price and avoid crime for the imminent destiny years. repeating this course of action.
The writer existing the effect of obstacles which consist of Procedure:
final consequences of outlier in facts preprocessing and
 Choose the value for K which stands for the
presented GA for outlier discovery in information
numeral of clusters.
preprocessing diploma [18]. The resulting table labels the
 Initialize the K cluster centers.
last diverse papers which might be completed on crime facts
mining and collaborative mastering .  Decide the division relationships of the N items by
conveying them to the adjoining cluster centre.
F. Criminal prediction using Naive Bayes theory  Re-estimate the K cluster centers, by assuming the
Mehmet Sait, and Mustafa Gok proposed the criminal memberships found over be acceptable.
prediction for finding the maximum viable crook of a  Replicate and until not any of the N items distorted
specific crime occurrence while the meant list of criminals membership in the preceding iteration.
are supplying with the criminal statistics that's brought on K-manner algorithm complexity where n is example, c is
synthetically the usage of Gaussian Mixture Model [19]. clusters and t is iterations and relatively gifted. It regularly
Numerous crimes performed via criminals and predict the terminates at a limited pinnacle-great. Its disadvantage is
unintended of each crime that may another time be germane simplest whilst mean is described and necessitate
completed through that criminal. The authors used Apriori to outline c , the range of clusters in development. It is not
exercise for commonplace object set organization that can capable to maintain deafening statistics and outliers and now
be accomplished with the aid of the criminals. not apt to find out clusters with non-convex shapes. K-
G. CRIMECAST: A Crime Prediction and Strategy means chosen to be used at this point because the
effortlessness of implementing it by means of python,
Direction Service however its effortlessness and rapidity which is dreadfully
Nafiz M., et al. Added CRIMECAST, a specific simulation pleasing in observe and it is fitting for elevated capacity
tool that examines beyond crime moves, styles, systems crime dataset and can facilitate to take out constructive
have an effect on crime, Crime incidence, crime occupied information.
vicinity, crime came about time, form of crime and useless Here we use Euclidean distance to find distance between
considering the fact that past crime facts as much as 30 two points. It measures the shortest distance between the
years to estimate destiny crime. Tahani A., et al. Studied two points. This distance is often used for data points in
two special crime records the usage of Decision Tree and Euclidean space. The Euclidean distance examines the root
Naïve Bayesian classifier to trace the maximum feasible of square difference coordinates of a pair of objects.
crime locations and their ordinary incidence time the usage
of Apriori Algorithm [20,21]. The authors provided what √∑
type of crime may occur subsequent in an exact location
within a sure time and mixing crimes dataset with its WCSS (within clusters of sum of squares) is defined as the
demographics. summation of the squared space among each component of
the cluster and its centriod. To compute WCSS, we require
to foremost find out Euclidean distance among cluster and
III. Enhanced k-means clustering its centroid. Then iterate this progression for all points in the
K-means clustering is one of the best, easy and most cluster, and then computation the values for the cluster and
important methods of the cluster examination. K-means segregate by the number of points [25]. As a final point,
clustering procedure is uncomplicated and it begins with a estimate the average diagonally the entire clusters.
portrayal of the algorithm .Foremost opt k preliminary Mathematically:
centriods, wherever k is a restraint which stands for the
numeral of the clusters. Each one spot is assigned to the WCSS = ∑
adjoining centriod and every assortment of points assigned
to a centriod is a huddle. The centriod of each group is then Crime evaluation is defined as diagnostic practices which
renew based on the points assigned to the cluster [22]. have enough money extensive records comparative to crime
Reiterate the obligation and renew steps awaiting no point styles and tendency correspondences to resource sensitive in

©Mody University of Science and Technology, Lakshmangarh 32

Mody University International Journal of Computing and Engineering Research
Volume 7 Issue 1, 2023, 31 - 34

scheduling and consumption of resources for the detection Idaho 14.2 54 120 2.6
and repression of criminal proceedings. It is imperative to Illinois 24 83 249 10.4
investigate crime because of following motives: Indiana 21 65 113 7.2
 Analyze crime to file to by way of regulation Iowa 11.3 57 56 2.2
enforcers about unambiguous crime traits in Kansas 18 66 115 6
appropriate technique. Kentucky 16.3 52 109 9.7
 Analyze crime to take advantage of the Louisiana 22.2 66 249 15.4
Maine 7.8 51 83 2.1
prosperity of information offered within the
impartiality system.
B. Implementation

 Here we positioned into motion the okay approach Set

of rules to preprocessed datasets.
 Take the popular propose cost from the clusters. Mean
value is based upon upon the amount of cluster to be
customary. Number of cluster to be produced might be
will be appropriately depends upon the range of imply
rate to be selected.
 Investigate for the adjoining mean value and situate in
cluster. Replicate the equivalent practice until we find
the similar cluster.
 And this process is implemented using python
programming language. And we use python in
PySpark tool.

Fig 1. Flow chart of crime detection

IV. RESULT AND DISCUSSION

A. Dataset
It is taken from the website of US crime price. Kaggle is one Fig3. Plot between WCSS and number of clusters
of the summit web web sites inside the floor of statistics
technological understanding which offers the whole-blown  Here we create a scree plot. It is a plot between
dataset. This dataset is composed of the niceties like WCSS(Within cluster sum of squares)and number of
Murder, Assault, Urban pop, Rape. And this dataset also clusters.
consists of the clusters of their relevant states of US states.  Without the domain knowledge or unclear motives,
we can not do anything.
TABLE I. Attributes of Crime Data  This plot helps us to decide the number of clusters to
specify.
Murder Assault Urban Rape
pop
Alabama 21.2 58 236 13.2
Alaska 44.5 48 263 10
Arizona 31 80 294 8.1
Arkansas 19.5 50 190 8.8
California 40.6 91 276 9
Colorado 38.7 78 204 7.9
Connecticut 11.1 77 110 3.3
Delaware 15.8 72 238 5.9
Florida 31.9 80 335 15.4
Georgia 25.8 60 211 17.4
Fig 4. Clusters between Murder and Rape
Hawaii 20.2 83 46 5.3

©Mody University of Science and Technology, Lakshmangarh 33

Mody University International Journal of Computing and Engineering Research
Volume 7 Issue 1, 2023, 31 - 34

 The beyond stature shows an allocation of how International Conference on rising Technological Trends [ICETT].
IEEE (2016).
states are sprinkled and clusters are noticeable based
on Murders and Rapes. [10] Alphonse Inbaraj, X., Rao, A.S.: Hybrid agglomeration algorithms for
 And here is an affirmative correlation between the crime pattern analysis. In: 2018 IEEE International Conference on
Current Trends toward affiliation Technologies, Coimbatore, India.
occurrences of Murder and Rapes in dissimilar
states. [11] Dutta, S., Gupta, A.K., Narayan, N.: Identity crime detection
Thus, the crime details give a perfect analysis for police exploitation technique. In: 2017 International Conference on
methodology Intelligence and Networks. IEEE (2017).
officials to take control of that particular area. By using
python programming language in PySpark tool we employ k [12] Manish Gupta, B. Chandra M.P. Gupta, ”Crime technique for Indian
police system”, Journal of Crime, Vol.2,No.6, 2006.
means clustering algorithm. And the accomplishment is
victorious for the data which have taken above. [13] David J. Hand, Heikki Mannila and Padhraic Smyth, “Principles of
knowledge mining”, MIT Press, 2001.

V. CONCLUSION AND FUTURE WORK [14] Hsinchun genus, Wingyan Chung, Yi Qin, archangel Chau, Jennifer
Jie Xu, Gang Wang, Rong Zheng, Homa Atabakhsh, “Crime
information Mining: A General Framework and sort of Examples”,
As of now, the project relies on handbook contribution from IEEE notebook computer Society Apr 2004
human in order to go through particulars in the database. If
[15] Cattleman Barnadas, M. (2016). Machine learning applied to crime
we can make this centralized organization and unite it to all
prediction (Bachelor's thesis, Universitat Politècnica Delaware
the police stations nationwide and formulate FIR reporting Catalunya).
digital, then it would be fairly easier to envisage crimes in
[16] Pandas Module in Python, Retrieved from http://pandas.pydata.org/
that fastidious locality and be acquainted with patterns in
them. It would also persuade residents to follow their [17] P. Thongtae and S. Srisuk, “An analysis of knowledge mining
niceties online. We can also evade bribery as the supervision applications in crime domain”, IEEE eighth International Conference
on notebook computer and IT Workshops, 2008
can keep a way on the quantity of cases registered and their
solvability rate which can assist them exploit their [18] Lawrence McClendon and Natarajan Meghanathan, “Using Machine
Learning rule To Analyze Crime Data”, a world Journal (MLAIJ)
possessions superior. 2015.
One of the maximum crucial troubles that have to be
addressed in the model provided in this task to improve the [19] Anushka Kumar, Vishnudas S, R. Kayalvizhi, " Using Map reduce
Techniques to Predict and Examine Crime Pattern", In-ternational
clustering procedure and crime detection is the optimization Journal of Engineering & Technology, 2018.
of the quantity of clusters within the clustering method and
the optimization of the techniques used in the prediction [20] Abba Babakura, Md Nasir Sulaiman and Mahmud A. Yusuf,
"Improved methodology of Classification rule for Crime Prediction",
section of the version development. In one word, we intend International conference on statistics and Security Technologies
to improve this project and to solve the prevailing (ISBAST), 2014
boundaries of the modern procedures to obtain best [21] Andrey Bogomolov A, Lepri B, Staiano J, Oliver N, Pianesi F,
outcomes and better performance. Pentland A, "Once upon a crime: towards crime prediction from
demographics and mobile data", 2014 New vogue month twelve (pp.
REFERENCES 427-434). ACM.
[22] F. Ozgul, C. Atzenbeck, A. Çelik and Z. Erdem, Incorporating
[1] Priyanka Gera and Rajan vohra, town Crime identification information sources and techniqueologies for crime method, in
exploitation Cluster Analysis, International Journal of technology and Intelligence and Security science (ISI), 2011 IEEE Int. Conf. (IEEE,
information Technologies IJCSIT, Vol.5(4), 2014. Beijing, China, 2011), pp. 176–180.
[2] A. Malathi. S. Santhosh man, algorithmic Crime Prediction Model
supported the analysis of Crime Clusters, international Journal of [23] Y. Peng, G. Kou, Y. Shi and Z. Chen, A descriptive framework for the
technology and Technology Volume eleven Issue eleven Version eld of {of information of data of information mining and knowledge
one.0July2011. discovery, In-ternational Journal of information Technology &
deciding 7(4) (2008) 639–682.
[3] Chris Delaney, Crime Pattern Defintions for study Analysis,
International Association of Crime Analysts (IACA), August 2011. [24] R. B. Santos, Electiveness of police in reducing crimes then the role of
[4] Jyothi Agarwal, RenukaNagal, Rajni Seghal, “Crime Analysis crime analysis, in Crime Analysis with Crime Mapping, ed. R. B. city
exploitation K-Means Clustering”, International Journal of private (Sage, California, 2012), pp. 40–53.
laptop Applications, Vol.83, No.04, 2013.
[5] Manish Gupta, B. Chandra and M.P. Gupta, “Crime technique for [25] Miss McCue, “Data Mining and revelatory Analysis: Intelligence
Indian Police information System”, notebook computer Society of Gathering and Crime Analysis”, Butterworth Heinemann, 2014.
land, Vol.40, No.1, 2008.
[6] K.Bogahawatte, and S.Adikari,”Intelligent criminal identification
system”, IEEE 2013 the eighth International Conference on
technology & Education (ICCSE) 2013 Apr twenty eight, 2013.
[7] Shermila, M.A., Bella mine, A.B., Santiago, N.: Crime information
analysis and prediction of soul identity exploitation machine learning
approach. In: 2018 ordinal International Conference on Trends in
subject and science (ICOEI2018). IEEE (2018).
[8] Yadav, S., Timbadia, M., Yadav, A., Vishwakarma, R., Yadav, N.:
Crime pattern detection, analysis and prediction. In: 2017
International Conference on subject, Communication and nil.5
Technology ICECA 2017. IEEE (2017).
[9] Sivaranjani, S., Sivakumari, S., Aasha, M.: Crime prediction &
foretelling in Madras exploitation agglomeration approaches. In: 2016

©Mody University of Science and Technology, Lakshmangarh 34