Saroj et al | IJCSET(www.ijcset.

net) | July 2016 | Vol 6, Issue 7, 279-281

Review: Study on Simple k Mean and Modified K

Mean Clustering Technique
Saroj Kavita
Student of Masters of Technology, Assistant Professor,
Department of Computer Science and Engineering Department of Computer Science and Engineering
JCDM college of Engineering, SIRSA, GJU, Hisar, JCDM college of Engineering, SIRSA, GJU, Hisar,
Haryana, India Haryana, India

Abstract: The main aim of this review paper is to provide various data mining problems successfully. In this paper,
comprehensive review of simple k mean clustering and clustering analysis is done by using simple k mean
modified k mean clustering techniques. Clustering is used as clustering and Modified k mean clustering. Normalization
active research in various fields like statistics, pattern and Indexing is an important preprocessing step in to
recognition and machine learning etc. Cluster Analysis is data
mining tool for a large and multivariate database. Clustering
standardize the values of all variables from dynamic range
is the one of data mining techniques in which data is divided into specific range. Cluster analysis is type data mining
into the groups of similar objects and dissimilar objects into technique which is used to find data segmentation and
another group. Clustering is a suited example of unsupervised pattern information. By clustering the data people get the
classification. data distribution, observe the character of each cluster, and
make further study on particular clusters. The aim of cluster
Keywords: Clustering, clustering Technique: Simple K Mean analysis is that the objects in a group should be similar to
and Modified k mean clustering. one another and different from the objects in other groups.
Clustering is much better when there is greater similarity
within a group and greater the difference between the
I. INTRODUCTION groups. So we can say that raw data has to be used with the
Due to the increased availability of computer hardware and algorithm to extract useful information from it. Various
software and the rapid computerization of business, large clustering algorithms according to different techniques
amount of data has been collected and stored in databases. have been designed and applied to various data mining
Researchers have estimated that amount of information in problems successfully. The most commonly used
the world doubles for every 20 months [1]. However raw algorithms in Clustering are Hierarchical, Partitioning,
data cannot be used directly. Its real value is predicted by Density Based and Grid based algorithms. The popular
extracting information useful for decision support. In most clustering techniques which have been suggested so far are
areas, data analysis was traditionally a manual process. either partition based clustering or hierarchical clustering
When the size of data manipulation and exploration goes but both approaches have their own advantages and
beyond human capabilities, people look for computing disadvantages in terms of the number of clusters, shape of
technologies to automate the process. Data mining is one of clusters, and cluster overlapping. When any clustering
the youngest research activities in the field of computing algorithm is applied to the raw data, only then we can get
science and is defined as extraction of interesting (non- clusters which are useful as shown in fig. [2].
trivial, implicit, previously unknown and potentially useful)
patterns or knowledge from huge amount of data. Data
mining is the process of analyzing data from different
perspectives and summarizing it into useful information
[2]. Data mining consists of extract, transform, and load
transaction data onto the data warehouse system, Data
mining includes the anomaly detection, association rule
learning, classification, regression, summarization and
clustering. Data mining is one of the most important
research fields which are due to the expansion of both
computer hardware and software technologies, which has
imposed organizations to depend heavily on these
technologies. Data mining concepts and methods can be
applied in various fields like marketing, medicine, real
estate, customer relationship management, engineering,
web mining, etc. Various clustering algorithms according
to different techniques have been designed and applied to Fig 1: Stages of Clustering [3]

279
 Saroj et al | IJCSET(www.ijcset.net) | July 2016 | Vol 6, Issue 7, 279-281

II. PARTITIONING CLUSTERING made of a number of attributes, any of which can be

Data objects are partitioned into non overlapping clusters nominal, numeric or a string. Data set is used here medical
so that each and every object is in exactly in one subset. data.
The reason of division of the data into several subsets is
that checking of the all possible subset systems is 3) Overview of WEKA Tool
computationally not feasible; there are certain greedy
For working with WEKA we do not have any need the
heuristics schemes which are used in the form of iterative
deep knowledge of data mining because it is very popular
optimization. This means different relocation schemes that
data mining tool. WEKA is open source and freely
iteratively reassign points between the k clusters [3].
available as well as platform-independent user and provides
many facilities. It also provides the graphical user interface
to the user. WEKA is a landmark system in the history of
the data mining and machine learning research
communities, because it is the only toolkit that has gained
such widespread adoption and survived for an extended
period of time. It provides different algorithms for data
mining and machine learning. Simple k mean clustering is
Fig 2 before and after partitioning
done by using this tool. We have to give the different
datasets as input to the WEKA tool. After that it will
A. Simple K-Means Clustering process the input and gives the output. In this way clusters
It is a partitioning method which finds mutual exclusive will be formed.
clusters of spherical shape. It generates a specific number
of disjoint, flat (non-hierarchical) clusters. K-Means
algorithm recognizes objects into k – partitions where each
partition represents a cluster. We start out with initial set of
means and classify cases based on their distances to their
centers. Next, we compute the cluster means again, using
the cases that are assigned to the clusters; then, we
reclassify all cases based on the new set of means. We keep
repeating this step until cluster means don’t change
between successive steps. Finally, we calculate the means
of cluster once again and assign the cases to their
permanent clusters. [4]

1) Method for Simple K means clustering

1 Input: k= no. of clusters. D= data set that contains n
objects.
2 Output: Set of k clusters.
Fig 3 Front view of WEKA tool
Method:
1. Randomly choose k objects from D as the initial cluster
III. MODIFIED K MEAN CLUSTERING
centre.
Modified K Mean approach is designed to improve the
2. Repeat.
time, no. of iterations and sum of squared errors and this
3. Reassign each object to the cluster to which the object is provides much better result as compare to Simple K Mean
most similar, based on the mean value of the objects clustering done by using K Mean tool. This is simple to use
in the cluster. and it also provides graphical user interface to the user. In
4. Update the cluster means, i.e. calculate the mean value of Modified K Mean clustering data is reduced by normalized
the objects for each cluster. method and then the parameters such as time taken, no. of
5. until no change. iterations, Sum of squared errors are improved by using
normalized technique. Normalization index method used
for modification. In this method input data has to be
2) Dataset minimized by using indexing so that we can get the raw
The dataset is set of data items and this is a very basic data in sequence after that we calculate the Euclidean
concept of machine learning. A dataset is equivalent to a distance between different clusters. After that
two-dimensional spreadsheet or database table. In WEKA, normalization is done to minimize the sum of squared
dataset is implemented by the weka.core.Instances class. A errors and no. of iteration resulting in less execution time to
dataset is a collection of examples; each example can be make the grouping of clusters.
taken from one of class weka.core.Instance. Each Instance

280
 Saroj et al | IJCSET(www.ijcset.net) | July 2016 | Vol 6, Issue 7, 279-281

REFERENCES
[1] Shraddha Shukla and Naganna S. “A Review on K-means Data
Clustering Approach” International Journal of Information &
Computation Technology, Volume 4, 2014
[2] S.Anupama Kumar and M. N. Vijayalakshmi “Relevance of data
mining techniques in editification sector”, International Journal of
Machine Learning and Computing, Volume 3, Issue 1, February
2013.
[3] Saroj, Tripti Chaudhary, “Study on Various Clustering Techniques”,
International Journal of Computer Science and Information
Technologies, Volume 6, Issue 3, 2015.
[4] Narender Kumar, Vishal Verma, Vipin Saxena “Cluster analysis in
data mining using k-means method”, International Journal of
Computer Applications, Volume 76, Issue.12, August 2013.
[5] Aastha Joshi, Rajneet Kaur “A Review: Comparative Study of
Various Clustering Techniques in Data Mining” International
Journal of Advanced Research in Computer Science and Software
Engineering, Volume 3, Issue 3, March 2013.
[6] Amandeep Kaur Mann & Navneet Kaur “Review paper on
Clustering Techniques”, Global Journal of Computer Science
and Technology Software and Data Engineering, Volume 13,
Issue 5, Year 2013.
[7] Bharat Chaudhary, Manan Parikh “A Comparative Study of
Clustering Algorithm using WEKA Tool”, International Journal of
Application or Innovation in Engineering and Management,
Volume 1, Issue 2, October 2012.
[8] Vaishali R. Patel, Rupa G. Mehta, “Clustering Algorithms: A
Comprehensive Survey”, International Conference on Electronics,
Information and Communication Systems Engineering, 2011.

Fig 4 Flow chart of Modified k means clustering

First step according to the given flow chart is that we have

to provide the raw data for conversion of data using
indexing and calculation of Euclidean distance is done so
that grouping of clusters is done in similar and dissimilar
categories.

IV. CONCLUSION
K Mean clustering is the most important type of
Partitioning clustering. Partitioning clustering is the one in
which clusters are partitioned according to their distances.
K mean clustering is that technique in which K cluster is
chosen and cluster which are at the less distance from k
cluster is selected in one group and others which are
farthest from the K cluster is placed in different group. In
this paper simple k mean clustering has been described by
using the WEKA tool and taking the medical data set i.e
Pima, Aids, Breast Cancer. On the other hand Modified k
mean clustering has been described based on normalization
and indexing approach using .NET which takes less time
with minimum no. of sum of squared errors to execute the
cluster.

ACKNOWLEDGMENT
I would like to thank Computer Science Engineering
department of JCDMCOE, SIRSA for the support and
providing an environment for this research work.

281