An Intelligent System A Comparative Study Of Fuzzy C-Means And K-Means Clustering Techniques

Afirah, Taufik (2013) An Intelligent System A Comparative Study Of Fuzzy C-Means And K-Means Clustering Techniques. (Submitted)

[img] PDF (24 Pages)
An_Intelligent_System_A_Comparative_Study_Of_Fuzzy_C-Means_And_K-Means_Clustering_Techniques.pdf - Submitted Version

Download (258kB)
[img] PDF (Full Text)
An_Intelligent_System_A_Comparative_Study_Of_Fuzzy_C-Means_And_K-Means_Clustering_Techniques.pdf - Submitted Version
Restricted to Registered users only

Download (702kB)


Clustering analysis has been considered as useful means for identifying patterns of dataset. The aim for this analysis is to decide what is the most suitable algorithm to be used when dealings with new scatter data. In this analysis, two important clustering algorithms namely fuzzy c-means and k-means clustering algorithms are compared. These algorithms are applied to synthetic data 2-dimensional dataset. The numbers of data points as well as the number of clusters are determined, with that the behavior patterns of both the algorithm are analyzed. Quality of clustering is based on lowest distance and highest membership similarity between the points and the centre cluster in one cluster, known as inter-class cluster similarity. Fuzzy c-means and k-means clustering are compared based on the inter-class cluster similarity by obtaining the minimum value of summation of distance. Additionally, in fuzzy c-means algorithm, most researchers fix weighting exponent (m) to a conventional value of 2 which might not be the appropriate for all applications. In order to find m, also called as fuzziness coefficient, optimal in fuzzy c-means on particular dataset is based on minimal reconstruction error.

Item Type: Article
Uncontrolled Keywords: Fuzzy logic, Fuzzy sets
Subjects: Q Science > Q Science (General)
Q Science > QA Mathematics
Divisions: Library > Projek Sarjana Muda > FTMK
Depositing User: Jefridzain Jaafar
Date Deposited: 27 Jan 2015 00:21
Last Modified: 28 May 2015 04:34
Statistic Details: View Download Statistic

Actions (login required)

View Item View Item