摘要如今的社会在经济迅速发展的作用下，推动了商业的发展。那么商业的发展崛起必定会产生大量的商业用户的数据信息，而这些信息的种类是多样化的、量是庞大的、处理过程也是复杂的。如何利用这些商业用户的数据信息，通过数据挖掘技术对用户偏好进行挖掘，实现数据蕴藏的商业价值是现在企业首要关心的问题。所以对商业用户数据信息挖掘用户的偏好显得尤为重要，意义也非常重大。81909

挖掘出商业用户的偏好，也就意味着知道了用户的需求，根据挖掘出用户偏好需求可以对症下药。如何对症下药主要体现在对用户服务和营销，在了解了用户的偏好需求情况下，针对不同用户偏好需求的群体制定精准营销方案来产生利益，针对不同的用户偏好群体在服务方面实现个性化，提高用户的服务体验。

本文主要探讨了面向用户偏好挖掘的商业数据分析。首先，本文对一个跨国非商店在线数据基本信息处理和聚类算法研究。挖掘出用户购买最多的商品和光临商店最多国家，以及商品购买数量和商品单价的偏好。其次对银行营销的数据做分类决策树研究。对训练集建立模型一并且优化模型，优化模型后对测试集进行预测评估，然后除去一些关键影响因素建立新的模型，优化新模型二并预测评估，对比模型一优化前后和模型二预测错误率从而说明模型的有效性。从而挖掘客户是否订阅定期存款的偏好。最后对R语言包中的杂货店数据进行关联算法研究用户偏好行为。用支持度、置信度和提升度对数据进行规则挖掘和搜索其他的规则挖掘用户的偏好，并给出一些销售建议。本文全部过程使用数据挖掘工具R语言进行研究用户偏好行为。挖掘过程中包括了建立模型、分析模型并且评价解释模型，可视化呈现模型分析结果，以及有效的预测评估。

毕业论文关键词：用户偏好；数据挖掘；Apriori算法；K-均值算法；决策树

Business data analysis for user preference mining

Abstract Today's society in the role of rapid economic development, and promote the development of the business。 Then the rise of commercial development will generate a large number of commercial users of the data, and the types of information is perse, the amount is huge, the process is complex。 How to make use of the data information of these commercial users, through data mining technology to mining user preferences, to realize the commercial value of the data is the main concern of enterprises。 So it is very important and significant to mining user's preference of data information of commercial users。

Dig out the commercial user preferences, also means knowing the needs of users through mining user preferences can suit the remedy to the case。 How to suit the remedy to the case is mainly reflected in the services to users and marketing, in the understanding of the user's needs and preferences, according to the needs of different user preference groups make precision marketing scheme to generate interest, for different groups of user preferences in terms of service realization of personalized, high user service experience。

This paper mainly discusses the business data analysis of user preference mining。 First of all, this paper studies the basic information processing and clustering algorithm for an international non store online data base。 Mining users to buy the most goods and visit the largest number of stores in the country, as well as the number of goods and the price of the purchase price preference。 Second, the bank marketing data to do classification decision tree research。 The training set model and optimization model, optimization model on the test set predictions were evaluated, then removed some key influence factors of establishing new model, optimization model and evaluate and contrast model optimization before and model prediction error rate to illustrate the model effectiveness。 To explore whether customers subscribe to the preference of regular deposits。 Finally, the study of the relationship between the data of the R language in the grocery store data is studied。 With support, confidence and the degree of improvement of the data mining rules and search for other rules mining users' preferences, and give some sales proposals。 In this paper, we use the data mining tool R language to study user's preference behavior。 The mining process includes the establishment of the model, the analysis model and evaluation and interpretation model, visual presentation model analysis results, as well as effective prediction evaluation。