مطالعه موردی استفاده از تکنیک های داده کاوی در تجزیه و تحلیل ساز و برگ ارزش مشتری
|کد مقاله||سال انتشار||مقاله انگلیسی||ترجمه فارسی||تعداد کلمات|
|2588||2009||7 صفحه PDF||سفارش دهید||6200 کلمه|
Publisher : Elsevier - Science Direct (الزویر - ساینس دایرکت)
Journal : Expert Systems with Applications, Volume 36, Issue 3, Part 2, April 2009, Pages 5909–5915
This study applies K-means method, fuzzy c-means clustering method and bagged clustering algorithm to the analysis of customer value for an outfitter in Taipei, Taiwan. These three techniques bear similar philosophy for data classification. Thus, it would be of interest to know which clustering technique performs best in a real world case of evaluating customer value. Using cluster quality assessment, this study concludes that bagged clustering algorithm outperforms the other two methods. To conclude the analyses, this study also suggests marketing strategies for each cluster based on the results generated by bagged clustering technique.
Outdoor activities are gaining their popularity in Taiwan and opportunities for selling outdoor outfits are abundant. According to the Commerce Industrial Services Portal of the Ministry of Economic Affairs, Republic of China (http://gcis.nat.gov.tw/English/index.jsp), there were 52 outfitters, most of which were located in Northern and Central Taiwan areas at the end of 2006. The large number of outfitters brought competition, which led to the decrease of profits. To survive competition and sustain profits, the outfitter must identify and retain customers of high value and profit potentials. Achieving the aforementioned goals will require the outfitter to customize marketing strategies and fulfill the needs of different customers and also to allocate resources effectively and efficiently, based on a well-managed customer database. Managing customer database is not an easy task. As the transaction record of a company becomes much larger in size as the time goes by, it might be necessary to divide all customers into appropriate number of clusters based on some similarities in these customers by using data mining techniques, particularly the clustering techniques. The values of different customer groups can then be calculated and evaluated to provide useful decisional information for management to utilize resources rationally. A variety of clustering techniques is commonly seen in practice. This study will discuss three clustering techniques, i.e., K-means method, fuzzy c-means method, and bagged clustering algorithm. These three techniques bear similar philosophy. K-means method is the most commonly seen approach for classification (Davidson, 2002). Fuzzy c-means method, very similar to the philosophy of K-means method, uses membership grades for data clustering (Jain, Murty, & Flynn, 1999). Bagged clustering algorithm based on K-means method and hierarchical methods provides another way for clustering (Dolnicar & Leisch, 2004). Due to the similarities in philosophy, it would be of interest to use these three methods in a case study and then evaluate which clustering technique performs better under the same circumstances.
نتیجه گیری انگلیسی
This study applies three clustering techniques in grouping an outfitter’s customers. The first approach is K-means method, which is very easy to converge and has been used on large data set. The second approach is fuzzy c-means method, which uses membership grades for data clustering. The third approach is bagged clustering algorithm, which is a combination of partitioning methods and hierarchical methods to provide another way for data classification. K-means method can be viewed as the fundamental clustering method, while fuzzy c-means method and bagged clustering algorithm can be regarded as improved K-means method. The data set utilized in this study consists of the transaction record from 551 customers shopping at the outfit store in Taipei, Taiwan from April 24 to March 26. The profile for each customer includes the membership number, gender, birth date, zip code, shopping frequency, and the total spending at the store. Six clusters are formed for each method. Then two types of cluster quality assessment are performed. The results show that bagged clustering algorithm is the best among the three techniques discussed here. Subsequently, marketing strategies are suggested in accordance with the results generated by bagged clustering algorithm. Cluster 6 is the big spender and the outfitter should try to attract these customers to shop at the store more often. Cluster 2 is the outfitter’s core customer. The outfitter should keep frequent contacts with these members, maintain their attitudinal and behavioral loyalty and encourage these customers to spend more. Cluster 5 is defined as frequent shoppers with the highest shopping frequency but low average total spending and low average spending per visit. These customers might have attitudinal loyalty but still lack behavioral loyalty. The outfitter can encourage these customers to spend more by using sales promotion, cross-selling, or up-selling. The customers in Cluster 1 are uncertain shoppers, which are all female from the North; these customers did not provide significant contributions to the outfitter. The outfitter should investigate whether these customers are new or have special preference for outdoor outfits by checking recency data and the merchandise these customers purchased. Customers in Cluster 1 who do not belong to these two types should be ignored to save marketing resources.