An Improved Bee Colony Optimization Algorithm for Clustering Analysis

• Main Author: 蔡政軒
• Other Authors: 駱至中
• Format: Others
• Language: zh-TW
• Published: 2011
• Online Access: http://ndltd.ncl.edu.tw/handle/58641926785303298424

碩士 === 佛光大學 === 資訊學系 === 99 === Clustering analysis is a popular data analysis and/or data mining task. In recent years, many nature inspired metaheuristic algorithms were developed to help clustering data into groups for further usages or follow-up surveys. Bee Colony Optimization (BCO) algorithm which is a recent proposed nature inspired metaheuristic algorithm for optimization, simulates the intelligent foraging behavior of honey bee swarms. In this thesis, an improved version of BCO is proposed to group objects into multiple clusters based on their optimal similarities. Furthermore, the improved BCO algorithm is also utilized to perform a series of clustering analysis that help finding suitable number of clusters. Performance of the proposed BCO algorithm with different objective functions and selection strategies have all been evaluated by using data sets from the UCI Machine Learning Repository and benchmarked with other popular heuristic algorithms. The experimental results show that the proposed BCO based clustering algorithm performs better and faster than other metaheuristic based approaches. In the meantime, the experimental results also indicates that the proposed BCO based clustering algorithm can provide suggestions that help deciding the number of clusters in order to get better clustering results.