Applying K-Means Clustering Algorithm to Discover Knowledge from Insurance Dataset Using WEKA Tool

The International Journal Of Engineering And Science (IJES)
|| Volume || 5 || Issue || 10 || Pages || PP 35-39 || 2016 ||
ISSN (e): 2319 – 1813 ISSN (p): 2319 – 1805
www.theijes.com The IJES Page 35
Applying K-Means Clustering Algorithm to Discover Knowledge
from Insurance Dataset Using WEKA Tool
Dr. Abdelrahman Elsharif Karrar1
, Marwa Abdelhameed Abdalrahman2
,
Moez Mutasim Ali3
1
College of Computer Science and Engineering, Taibah University, Saudi Arabia
2
College of Computer Science and Information Technology, University of Science and Technology, Sudan
3
College of Computer Science and Information Technology, University of Science and Technology, Sudan
--------------------------------------------------------ABSTRACT-------------------------------------------------------------
Data mining works to extract information known in advance from the enormous quantities of data which can
lead to knowledge. It provides information that helps to make good decisions. The effectiveness of data mining
in access to knowledge to achieve the goal of which is the discovery of the hidden facts contained in databases
and through the use of multiple technologies. Clustering is organizing data into clusters or groups such that
they have high intra-cluster similarity and low inter cluster similarity. This paper deals with K-means clustering
algorithm which collect a number of data based on the characteristics and attributes of this data, and process
the Clustering by reducing the distances between the data center. This algorithm is applied using open source
tool called WEKA, with the Insurance dataset as its input.
Keywords: Clustering, Centroid, Data Mining, knowledge discovery, K-means, WEKA.
---------------------------------------------------------------------------------------------------------------------------------------
Date of Submission: 17 May 2016 Date of Accepted: 05 November 2016
--------------------------------------------------------------------------------------------------------------------------------------
I. INTRODUCTION
The present age is characterized by the use of advanced data technology to save and retrieve data and enormous
quantities and which is described data warehousing. These data provide open the door to a range of specialized
in the management of those data subjects was the most prominent topic of data mining, which is one of the
important methods to get useful information from the data. Scientists have known the term data mining as "part
of the process of knowledge discovery in databases, which are made using multiple methods its goal configure
models of data". [1]
Data mining techniques are designed to extract hidden information in the database, this modern technology has
imposed itself firmly in the information age and in the light of the great technological development and
widespread use of databases, they offer institutions in all areas the ability to explore and focus on the most
important information in the databases. Data mining techniques focus on predictions future, explore behavior
and trends allowing to take decisions correct and taken at the right time, also as the data mining techniques to
answer Many questions in record time, especially those questions that are difficult to answer them by using
methods statistical traditional. [2]
II. DATA MINING
Data mining technology is simply extract the important information from huge amount of information to follow
certain mechanisms analysis this information, or it’s a technique used in the process of extracting data from data
warehouses. Data Mining passes a number of stages starting from data cleansing, and standardization of data,
and the relevant test data, then transfer them, classify then evaluated and extract data. [3]
There are many definitions of the concept of data mining:
 Is a computerized search for the knowledge of data without prior assumptions about what can be this
knowledge?
 Is the process used by companies to convert the raw data into useful information?
 Analysis of large sets of data and summarize the data in the new forms be understandable and useful to its
users and is done in banks telecommunications companies commercial transactions and scientific data
(Biology - Astronomy).
 Process data analysis by linking them with artificial intelligence techniques and statistical process, is simply a
process of exploration and search for specific and useful information in a huge amount of data.

Applying K-means Clustering Algorithm to Discover Knowledge from Insurance Dataset Using…
 Search the relevant information together collected by common characteristics and linked unit subject or
specialization is then to find this information between a very large amounts of information that has no
relationship was presented to the decision maker. [3]
III. TECHNIQUES OR METHODS OF DATA MINING
Data mining process uses several techniques through which you can discover the hidden trends and models in
large amounts of data and can be used one or more of these techniques are as follows:
 Prediction: Use of available data and the application of certain techniques to give them values successful
future.
 Description: Process description available data to see their ratings by the presence of the relationships
between them.
 Classification: It is a set of data analysis to create a set of database assembled that can be used to classify
any future data to find information that relates to the common characteristics and classification many tools
such as decision tree, nearest neighbor and regression.
 Association: Is the database that includes fixed coupling relationships among a group of objects in the
database any association between the occurrence of an event and another event occurs, which is often
called the market basket analysis.
 Sequential Analysis: Which is similar to association and placed in the name the link, but analysis linked in
time in the Search for models occurs in any deal with the succession of data that occur in separate cases.
 Clustering: The idea of collecting data is a simple idea in nature and very close to the human way of
thinking where we are whenever we deal with a large amount of data tends to summarize the vast amount
of data into a small number of groups or categories in order to facilitate the process of analysis. Algorithms
assembly used widely not only for the organization and classification of data but also the data compression
and build a model arrangement.The process of clustering is the process of collecting objects or items that
possess the qualities and attributes are similar in groups called clusters. The process of clustering one of the
main roads in the process of data mining and can be used as a standalone tool to gain insight into how the
distribution of data, control characteristics of each group and focus on a specific set of groups so for further
analysis and can be as a preliminary step to the work of an elementary or other techniques such as
characterization and classification. [4]
IV. K-MEANS ALGORITHM
Is one of the clustering algorithms, it collect a number of data based on the characteristics and attributes of this
data and the process of the Clustering by reducing the distances between the data center. The steps of this
algorithm are:
 Determine the number of clusters K which is a step Initialize Preliminary.
 Determine the coordinates of the centers of clusters (Centroid) randomly for the first time and calculate the
average of the points that belong to the center for the rest of the times.
 Calculate the distance between each example and among all centers and is used Euclidean dimension. Given
the Euclidean distance ) ij
d
) between the two examples( i,, j ) the following relationship:
 Data collection (examples) with its nearest center.
 Repeat steps 2 through 4 until you get stability (and the absence of moving objects within the clusters), or
even repeating a certain number of times. [5]
Before you start in the presentation of the results of the application must describe the data which were used in
the search: this paper apply the algorithm (k-means) to the insurance company data to clarify the best payment
method that can be available to the customers, as the company suffered from the challenges and difficulties
limit their ability to deal with payment methods. This was done using a program WEKA.
V. WEKA
WEKA is a data mining system developed by the University of Waikato in New Zealand that implements data
mining algorithms using the JAVA language. WEKA is a state-ofthe-art facility for developing machine
learning (ML) techniques and their application to real-world data mining problems. It is a collection of machine
learning algorithms for data mining tasks. The algorithms are applied directly to a dataset. [6]



n
k
jkikij
xxd
1
2
)(

WEKA implements algorithms for data preprocessing, classification, regression, clustering and association
rules. It also includes visualization tools. [6]
To perform cluster analysis in weka the dataset is needed to be loaded to weka and it should be in the format of
CSV or ARFF file format. If the dataset is not in arff format we need to be converting it.
Figure 1: WEKA GUI
Figure 2: Customer Data in the Insurance Company

Figure 3: Choose the Data File
Figure 4: Choosing the Attributes
Figure 5: Cluster mode

Figure 6: Clustered output
VI. RESULT DISCUSSION
When we study the output report we can see four clusters (cluster 0, cluster1, cluster2, cluster3). In the report
appear as:
Number of clusters selected by cross validation: 4
The first column gives you the overall population centroid. The second and third and fourth and fifth columns
give you the centroids for cluster 0, 1, 2 and 3 respectively. Each row gives the centroid coordinate for the
specific dimension.
Here we must notice that finding the centroids is an essential part of the algorithm. The centroids are a result of
a specific run of the algorithm and are not unique, a different run may generate a different centroid set.
For each cluster there is “clusters prior” probability. The estimators consist of a number for each possible
attribute value, and the attribute values are treated in order.
 Cluster0 has total 4 objects, out of which majority of objects (6) data sets.
 Cluster1 has total of 22 objects, out of which majority of objects (32) data sets.
VII. CONCLUSION
Clustering is organizing data into clusters or groups such that they have high intra-cluster similarity and low
inter cluster similarity. K-means algorithm is one of the clustering algorithms, it collects a number of data based
on their characteristics and attributes, and run the process of Clustering by reducing the distances between the
data center. WEKA an open source tool is used to apply K-means algorithm on insurance dataset.
REFERENCE
[1]. Daniel T. Larose, “Data Mining Methods and Models”, 2011.
[2]. G. K. Gupta, “Introduction to Data Mining with Case Studies”, 2006.
[3]. Gregory Piatetsky, “From Data Mining to Knowledge Discovery:An Introduction”, 2012.
[4]. Jiawei Han, Micheline Kamber, “Data Mining: Concepts and Techniques (Second Edition)”, 2014.
[5]. Subhash Sharma, Ajith Kumar, “Cluster Analysis and Factor Analysis”, 2014.
[6]. Yizhou Sun, “an Introduction to WEKA”, 2008.

Applying K-Means Clustering Algorithm to Discover Knowledge from Insurance Dataset Using WEKA Tool

Recommended

More Related Content

What's hot (19)

Viewers also liked (8)

Similar to Applying K-Means Clustering Algorithm to Discover Knowledge from Insurance Dataset Using WEKA Tool (19)

Recently uploaded (20)

Applying K-Means Clustering Algorithm to Discover Knowledge from Insurance Dataset Using WEKA Tool