customer behavior:客户行为检测决策树数据分析Detecting the change of customer behavior based on decision tree analysis [11]
论文作者:留学生论文论文属性:硕士毕业论文 thesis登出时间:2011-01-14编辑:anterran点击率:25052
论文字数:9758论文编号:org201101141122036887语种:英语 English地区:韩国价格:免费论文
关键词:data miningdecision treechange analysisInternet shopping mall
(96%rate of decrease) in sales for customers who are in their 40s,order frequently and visit the mall frequently. Because ofthe large rate of decrease, this segment of customers willdisappear in the near future. Therefore special carefulconsideration of these customers will be necessary. Fromchanged rule (2) in Table 3, we find a rapid growth ofcustomer groups with a high purchasing rate in sales forcustomers who are in their 20s and visit the mall frequently.This tendency is obviously desirable; therefore a marketingcampaign to encourage the revisiting of these customersshould be developed. With regard to the unexpectedchanges of Table 4, the sales for customers who are intheir teens and visit the mall frequently are low from thefirst data set. But in the second data set, we can see that theTable 2: Number of changed rules in the same attribute andthe same cut at time tType of change Number ofchangedrulesNumber of significantchanged rulesEmerging patterns 4 3 (degree of change >0.1)Unexpected changes 1 1 (degree of change >0.1)Added=Perished rules 1 1 (degree of change >0.001)Table 5: Significant perished rules (degree of change>0.001) in the same attribute and the same cut attime trtiMSV Support aij1 If Sales visit¼High,Age¼40s, Order count¼Low, then Sales amount¼High0 0.008 0.008Table 3: Significant emerging patterns (degree of change >0.1) in the same attribute and the same cut at time tNo. Rules supt(ri) suptþk(rj) aij1 If Sales visit¼High, Age¼40s, Order count¼High, then Sales amount¼High 0.09 0.00 0.95852 If Sales visit¼High, Age¼20s, then Sales amount¼High 0.18 0.22 0.21383 If Sales visit¼Low, Payment¼Card, then Sales amount¼Low 0.23 0.20 0.1048Table 4: Significant unexpected changes (degree of change>0.1) in the same attribute and the same cut attime trtirtþkj dij d0ij aij1 If Sales visit¼High, Age¼Teen,then Sales amountLowSupport¼0.003If Sales visit¼High,Age¼Teen, thenSales amount¼HighSupport¼0.0050.5 0.5 0.6Sales visitLow HighLowHighTotal56.4%43.6%100.0%334925945943LowHighTotal78.0%22.0%100.0%30388333871PaymentCash CardLowHighTotal63.2%36.8%100.0%6984071105Reserved MoneyHigh LowLowHighTotal68.3%31.7%100.0%1151534168586.3%13.7%100.0%1187299218678.1%21.9%100.0%45312758015.0%85.0%100.0%31117612072LowHighTotalLowHighTotalLowHighTotalFigure 5: Decision tree of the data set at time tþk in the same attribute and the same cut.Expert Systems, September 2005, Vol. 22, No. 4
sales for these customers are high. This means that theimportance of such customers is gradually increasing.Therefore, a modification of the existing marketingstrategy and plan for the customers is required. Finally,one perished rule is found in Table 5. From February toJune in 2000, the sales of frequently visiting customers, intheir 40s and who order infrequently were high. But afterJune 2000, we cannot find these rules anymore. Thereforewe should decide whether additional services and productsfor elders need to be developed.5.2. Same attribute and same cut at time tþkAs with the case of the same attribute and same cut at time t,we built the decision tree and rules of the data set at timetþk with the SAS 8.0 enterprise miner program. And withSales visitLow HighPaymentCash CardLowHighTotal75.4%24.6%100.0%510166676Reserved MoneyHigh LowLowHighTotal78.5%21.5%100.0%803219
本论文由英语论文网提供整理,提供论文代写,英语论文代写,代写论文,代写英语论文,代写留学生论文,代写英文论文,留学生论文代写相关核心关键词搜索。