本文選題：數(shù)據(jù)挖掘 + 關(guān)聯(lián)規(guī)則　； 參考：《鄭州大學》2013年碩士論文

【摘要】：如何將數(shù)據(jù)倉庫及數(shù)據(jù)挖掘的相關(guān)技術(shù)運用到銀行金融產(chǎn)品的規(guī)劃與銷售中,是目前我國金融行業(yè)較為迫切需要研究的領(lǐng)域。該領(lǐng)域研究的內(nèi)容包括數(shù)據(jù)挖掘技術(shù)的研究、更加有效的挖掘算法設(shè)計、客戶關(guān)系管理系統(tǒng)的重新構(gòu)建等方面。本文具體探討了銀行理財產(chǎn)品銷售分析系統(tǒng)在實施過程中的若干關(guān)鍵技術(shù),同時提出了一種有效挖掘負關(guān)聯(lián)規(guī)則的方法。 銀行客戶的眾多行為中,存在著正、負關(guān)聯(lián)規(guī)則。傳統(tǒng)的關(guān)聯(lián)規(guī)則算法僅反應(yīng)了正項之間的關(guān)聯(lián)關(guān)系,無法解決負關(guān)聯(lián)的問題。本文提出了一種有效的算法(GA_PNAR),用以解決銀行客戶行為負關(guān)聯(lián)的問題。GA_PNAR算法首先利用Apriori算法生成頻繁項集,之后利用基于相關(guān)系數(shù)的NRGA算法生成含有所有負項的關(guān)聯(lián)規(guī)則,在所有規(guī)則生成后,利用遺傳算法優(yōu)選生成的規(guī)則。GA_PNAR算法是一款非常有前途的發(fā)現(xiàn)規(guī)則的方法。 當前針對銀行營銷的方法主要考慮的是客戶的基本屬性,沒有全面考慮客戶的價值屬性以及行為屬性。營銷方案的設(shè)計也主要是對客戶進行細分,通過對客戶的基本屬性如投資期限、風險性偏好等進行聚類分析。根據(jù)聚類分析的結(jié)果,將客戶的聚類特征與理財產(chǎn)品的特征結(jié)合,為客戶提供理財方案。在這種方案中客戶的投資期限、風險性偏好等屬性往往通過測試獲得,存在很大的不準確性和失真。本文將GA_PNAR方法應(yīng)用于銀行客戶-產(chǎn)品之間關(guān)聯(lián)規(guī)則的發(fā)現(xiàn),該過程選取的客戶數(shù)據(jù)主要是客戶行為屬性。相比常規(guī)的聚類分析,客戶-理財產(chǎn)品關(guān)聯(lián)規(guī)則分析,能夠為銀行客戶提供更加精確、專業(yè)化的理財產(chǎn)品指導(dǎo)。另外,在該模型的數(shù)據(jù)預(yù)處理階段采用了云模型的方法對數(shù)值型屬性進行了概念化分層,該方法可以有效地解決數(shù)值分層的模糊性問題。最后,本文提出了一個銀行理財產(chǎn)品營銷系統(tǒng)的設(shè)計方案。 論文對于國內(nèi)金融行業(yè)實施結(jié)構(gòu)化數(shù)據(jù)挖掘技術(shù)、部署企業(yè)級數(shù)據(jù)倉庫、完善客戶分類、加強客戶關(guān)系管理、市場銷售分析、金融產(chǎn)品規(guī)劃、市場需求動態(tài)分析等各個方面均有一定的借鑒和現(xiàn)實指導(dǎo)意義。
[Abstract]:How to apply the related technology of data warehouse and data mining to the planning and sale of bank financial products is the most urgent need to be studied in our financial industry. The research in this field includes the research of data mining technology, the more effective mining algorithm, the re construction of the customer relationship management system and so on. In this paper, some key technologies in the implementation process of bank financial products sales analysis system are discussed, and an effective method for mining negative association rules is proposed.
In many behavior of bank customers, there are positive and negative association rules. The traditional association rule algorithm only reflects the relationship between positive items and can not solve the problem of negative correlation. This paper proposes an effective algorithm (GA_PNAR) to solve the negative association of bank customer behavior. The.GA_PNAR algorithm is first generated by the Apriori algorithm. Frequent itemsets, then use the NRGA algorithm based on correlation coefficients to generate association rules containing all negative items. After all rules are generated, the rule.GA_PNAR algorithm generated by genetic algorithm is a very promising method of discovering rules.
At present, the main consideration of the method of banking marketing is the basic attribute of the customer. It does not fully consider the value attribute and the behavior attribute of the customer. The design of the marketing plan is mainly to subdivide the customer, and through the clustering analysis of the basic attribute of the customer, such as the term of investment and the risk preference, etc., according to the result of the cluster analysis, The customer's clustering characteristics and the characteristics of financial products are combined to provide a financial plan for customers. In this scheme, the time limit of the customer's investment and the risk preference are often obtained by testing, and there is a lot of inaccuracy and distortion. This paper applies the GA_PNAR method to the discovery of Association rules between customers and products of the bank. Customer data are mainly customer behavior attributes. Compared to conventional clustering analysis, customer financial product association rules analysis can provide more accurate and professional financial product guidance for bank customers. In addition, a cloud model is used in the data preprocessing stage of the model to conceptualize the numerical attributes. The method can effectively solve the fuzziness of numerical layering. Finally, this paper proposes a design scheme for the marketing system of bank financial products.
This paper has a certain reference and practical significance for the domestic financial industry to implement structured data mining technology, deploy enterprise data warehouse, improve customer classification, strengthen customer relationship management, market sales analysis, financial product planning, market demand dynamic analysis and so on.

【學位授予單位】：鄭州大學
【學位級別】：碩士
【學位授予年份】：2013
【分類號】：TP311.13

【參考文獻】

相關(guān)期刊論文 前10條

1 董祥軍,王淑靜,宋瀚濤,陸玉昌;負關(guān)聯(lián)規(guī)則的研究[J];北京理工大學學報;2004年11期

2 陸安生,陳永強,屠浩文;決策樹C5算法的分析與應(yīng)用[J];電腦知識與技術(shù);2005年09期

3 盧雪燕;王勇;周永權(quán);;一種基于位矩陣的負關(guān)聯(lián)規(guī)則挖掘新算法[J];廣西民族大學學報(自然科學版);2007年04期

4 周艷平;顧幸生;;差分進化算法研究進展[J];化工自動化及儀表;2007年03期

5 陳明亮;客戶忠誠決定因素實證研究[J];管理科學學報;2003年05期

6 李德毅，孟海軍，，史雪梅;隸屬云和隸屬云發(fā)生器[J];計算機研究與發(fā)展;1995年06期

7 ;Mellon銀行預(yù)測數(shù)據(jù)挖掘前景光明[J];金卡工程;2003年09期

8 董祥軍,宋瀚濤,姜合,陸玉昌;基于最小興趣度的正、負關(guān)聯(lián)規(guī)則挖掘[J];計算機工程與應(yīng)用;2004年27期

9 董祥軍,陳建斌,宋麗哲,宋瀚濤,陸玉昌;基于定制時間約束的正負時態(tài)關(guān)聯(lián)規(guī)則挖掘[J];計算機工程與應(yīng)用;2004年28期

10 馬占欣;周文剛;陸玉昌;;基于最小提升率的正負關(guān)聯(lián)規(guī)則挖掘[J];計算機工程與應(yīng)用;2007年07期

相關(guān)博士學位論文 前1條

1 趙基;基于數(shù)據(jù)挖掘的銀行客戶分析管理關(guān)鍵技術(shù)研究[D];浙江大學;2005年

相關(guān)碩士學位論文 前5條

1 王春光;K-MEANS聚類算法在銀行個人客戶經(jīng)理管理中的應(yīng)用研究[D];內(nèi)蒙古大學;2011年

2 俞斌;改進K-MEANS聚類算法在銀行CRM中的應(yīng)用與研究[D];浙江工業(yè)大學;2011年

3 李娜;基于數(shù)據(jù)挖掘的零售客戶細分模型的應(yīng)用研究[D];四川師范大學;2008年

4 陳玲燕;基于遺傳算法的FCM聚類在銀行客戶細分中的應(yīng)用研究[D];蘭州商學院;2009年

5 游晉峰;含負項和數(shù)量屬性的關(guān)聯(lián)規(guī)則挖掘研究[D];四川師范大學;2010年

本文編號：1893176