Abstract:Evaluation of clustering validity to set up an optimal cluster-data space (CDS) is a vital task in many fields related to data mining. Almost existing clustering validity indexes (CVIs) lack stability due to being too sensitive to noise, especially impulse noise. Here, we (1) propose a new CVI named DzI (Dzung Index) or fRisk2 using analysis of fuzzy-set-based accumulated risk degree (FARD), and (2) present a new algorithm named fRisk2-bA for determining the optimal number of data clusters. It is a method of evaluation of the centroid-based fuzzy clustering validity. In essence, the fRisk2 still focuses on enhancing the data compression in each cluster and expanding the separation between cluster centroids. However, these features are exploited indirectly through FARD. As a result, the proposed method not only can avoid the difficulties of the traditional ones relying on the compression and separation properties directly but also can distill better local and global attributes in the data distribution to estimate the CDS more fully. Along with the proved theory basis, surveys, including the ones based on noisy datasets from measurements, showed the compared advantages of fRisk2 as follows. (1) The accuracy, stability, and convergence of the fRisk2 are outstanding. (2) Its total calculating cost is lower than the other surveyed CVIs.

Fuzziness Measurement of Fuzzy Sets and Its Application in Cluster Validity Analysis

A validity measure for fuzzy clustering

A kind of practical fuzzy clustering

A New Validity Index Based on Intra-Cluster Variation and Inter-Cluster Overlap

A Fuzzy Cluster Validity Index Based on Clustering Mistake Measures

A new validity measure for fuzzy c-means clustering

Cluster Validity for Fuzzy Criterion Clustering

Fuzzy C-Means Clustering Validity Function Based on Multiple Clustering Performance Evaluation Components

A New Validity Index Based on Fuzzy Energy and Fuzzy Entropy Measures in Fuzzy Clustering Problems

A New Validity Index for Fuzzy-Possibilistic C-Means Clustering

The interrelation in fuzzy clustering

The interelation in fuzzy clustering

Study on Hesitant Fuzzy Information Measures and Their Clustering Application

Adaptive Approach to Fuzzy Clustering

Fuzzy Partitional Clustering Algorithms

Novel Pythagorean fuzzy entropy and Pythagorean fuzzy cross-entropy measures and their applications

On the Index of Cluster Validity

A roughness measure of fuzzy sets from the perspective of distance

A Class of New Interval-Valued Intuitionistic Fuzzy Distance Measures and Their Applications in Discriminant Analysis

Distance-Based Knowledge Measure and Entropy for Interval-Valued Intuitionistic Fuzzy Sets

Centroid-based clustering validity: method and application to quantification of optimal cluster-data space