원문정보
초록
영어
In allusion to such problems as the no use of the structural information of the dataset in the traditional clustering effectiveness evaluation function and the excessive noisy point deletion, the research method integrating theoretical analysis and empirical analysis is adopted to establish KPI management index system model for telecommunication enterprises. A new clustering effectiveness evaluation function is proposed in this article. Specifically, PCA (principal component analysis) method in multivariate statistics is applied in the performance evaluation systems of telecommunication enterprises, and meanwhile relevant instances are analyzed and evaluated. Therein, the evaluation index system has the features of simpleness, strong practicability, low operation cost and high accuracy, and the geometric structure features of dataset are added for the performance evaluation of telecommunication enterprises. Additionally, distance critical value L is added in the compact indexes and the constraint condition thereof is also given in order to construct a new clustering effectiveness evaluation index model which can more scientifically and rationally reflect the actual evaluation result.
목차
1. Introduction
2. Big Data Clustering Effectiveness Evaluation Function
3. Construction of New Clustering Effectiveness Evaluation Function
3.1. Compactness Measurement
3.2. Separability Measurement Based on Dual Constraints to Noise Points
3.3. Construction of Clustering Effectiveness Function
4. Coalition Evaluation Method Based on Fuzzy Soft Set
4.1. Coalition Evaluation Process
5. Instance Analysis
6. Conclusion
References