A Case Study of Applying Data Mining Clustering Techniques in Customer Relationship Management
| Autor: | Yi-Ying Cheng, 鄭易英 |
|---|---|
| Rok vydání: | 2007 |
| Druh dokumentu: | 學位論文 ; thesis |
| Popis: | 95 This research applies data mining techniques in an outfitter in Taipei to analyze 551 customers’ transactions data. First, RFM model is used by constructing three two-dimensional distribution charts, namely F-R, M-R, and F-M charts, to identifying 23 valuable and potential valuable customers. Specifically, 18 out of 23 customers are male; 20 customers live in northern Taiwan; and 8 persons are in the age between 26 and 30.Only 5 customers belong to Water signs and each of the other three signs has 6 customers. To further analyze customers’ characteristics, this study based upon RFM model uses demographic variables to identify significant variables or interactions that have great impacts on recency, frequency, or monetary. Only gender, astrology, interaction between gender and area, and interaction between astrology and area are significant to monetary. However, there is no any significant factor or interaction to either recency or frequency. Thus, monetary, gender, astrology, and area are taken as variables along with the use of self-organizing map (SOM), two-step clustering technique, and K-means algorithm in clustering customers’ transactions data and in comparing the cluster quality among different clustering techniques. Based on cluster quality assessment, SOM performs best for this case study. Therefore, the clustering results generated by SOM can become the guidance for this outfitter to plan and implement marketing strategies. Additional investigations based upon the SOM clustering results are summarized as follows. There are 116 customers who spent more than 15,000 dollars, and the company should pay much attention to these customers. In contrast, there are 269 customers who spent less than 5,000 dollars, and the company, on the other hand, should not put too many resources on these customers. The company can develop specific marketing activities by applying different segment variables based on the cluster quality assessment to select the best clustering technique. |
| Databáze: | Networked Digital Library of Theses & Dissertations |
| Externí odkaz: |
|