符号序列聚类及在市场细分中的应用研究-计算机软件与理论专业论文.docx

英文摘要 英文摘要 ABSTRACT The dissertation is intended to provide consumable manufacturers an approach to understand the market segmentation structure, so as to create the new product strategy. The research is based on the Market Structure Analysis of Chang’an Vehicle Concept Design Evaluation System project. Nowadays, domestic auto-industry is experiencing the great shift from the stage of imitating its foreign counterparts to independently designing stage. And product concept design becomes a critical factor to the success of a new vehicle. However, the right decision in product concept design is driven by the accurate understanding of the competition situation and user requirement. That is the decision makers of such enterprises should give a clear answer that which products the consumers mostly prefer and which ones they dislike, and what are the most important features the new product should have and what are not from the consumers’ point of view, moreover, moreover, they should find reasons behind such phenomenon. In order to solve such a problem, the dissertation chooses the preference dataset, given by consumers, and segments the underline market, by adopting a data mining approach vs. the traditional statistics one. Following by this approach, a common framework to handle symbolic sequence clustering problems is proposed. The framework is consisted of four main parts: first of all, how to describe the similarity between different symbolic sequences, secondly, how to find clusters, then how to display such clusters in a easy way, and at last, why do such clusters exist. The framework was applied in the real world application and produced good results. The main contributes of this dissertation consist of the following two originalities: First, it segments the market based on consumers’ preference symbolic sequences other than the demographic data of consumers. Second, it proposed an overall research approach to symbolic sequence clustering problem, that i