運用資料採礦於銷售績效分析－以Clementine為例

此次研究將使用資料採礦的技術對維綸實業公司進行市場分析並進而做銷售產品評比。研究資料為維綸實業公司內部2008 至2010 所有銷售品項紀錄，其中包含各品項編號，對應的車型年份，所在的訂單編號，售價，銷售國家，毛利以及毛利率…等。以銷售國家為主要研究對象，將三年的毛利加總，平均毛利率和平均售價作為此國家的銷售特徵。利用集群分析將69 個國家依照相近的銷售特徵分為明星市場群（平均售價最高，總毛利最高，平均毛利最高），潛力市場群（總毛利次之）以及墊底市場群（總毛利最低）。為了更進一步了解此三大市場的銷售品項狀況，此研究利用RFM對各產品三年來的總毛利、購買時間點及購買頻率三項變數來計分，得出各市場高分的品項(購買頻率高、購買時間點離現今最近及毛利高)。最後根據80/20 法則，建議該公司應鞏固明星市場，深耕潛力市場（兩市場總毛利率佔三年來總毛利的80%以上）。

關鍵字

集群分析； Clementine ； RFM

並列摘要

In this study, we use technique of data mining to do market analysis and use the results to give sales recommendation. The research data is offer from auto parts industrial ltd., and the period is from 2008 to 2010. The data includes the product No., year of car style, order number, sales price, countries, profit, and average sales price. The goal mainly focus on countries the company contact with. We first take sum of profit, average of profit rate, and average sales price as the sale characteristic of countries. Next, we use cluster analysis to divide 69 countries to 3 groups, say ＂superstar market＂, ＂potential market＂, and ＂low-level market＂. Besides, in order to study sales situation of three groups, we use RFM to obtain the credit via total profit, time to buy product, and frequency of buying product. Finally, applying 80/20 rule, we suggest the company should keep contacting with superstar market, and develop the relation with potential market.