Open AccessDetection of Crimes Using Unsupervised Learning Techniques
Author(s) -
Rajesh Babu,
G. Snehal,
P. Aditya Satya Kiran
Publication year - 2017
Publication title -
aptikom journal on computer science and information technologies
Language(s) - English
Resource type - Journals
eISSN - 2528-2425
pISSN - 2528-2417
DOI - 10.11591/aptikom.j.csit.92
Subject(s) - cluster analysis , computer science , partition (number theory) , data mining , expectation–maximization algorithm , canopy clustering algorithm , partition problem , maximization , pattern recognition (psychology) , correlation clustering , artificial intelligence , mathematics , maximum likelihood , mathematical optimization , statistics , combinatorics
Data mining can be used to detect model crime problems. This paper is about the importance of data mining about its techniques and how we can easily solve the crime. Crime data will be stored in criminal’s database. To analyze the data easily we have data mining technique that is clustering. Clustering is a method to group identical characteristics in which the similarity is maximized or minimized. In clustering techniques also we have different type of algorithm, but in this paper we are using the k-means algorithm and expectation-maximization algorithm. We are using these techniques because these two techniques come under the partition algorithm. Partition algorithm is one of the best methods to solve crimes and to find the similar data and group it. K-means algorithm is used to partition the grouped object based on their means. Expectation-maximization algorithm is the extension of k-means algorithm here we partition the data based on their parameters.