應用資料探勘於IC 設計產業協助銷售策略-以X 公司為例

Abstract

高科技業市場變化快速，競爭壓力日益龐大，根據以往消費模式找出客戶可能的需求，並且快速反應客戶要求日益重要。 本研究以類比IC設計產業為例，應用資料探勘的技術協助銷售策略，透過客戶的RFM分數使用K-means進行分群，再使用關聯規則針對每個群集從歷史的銷售資料中找出客戶的購買模式，並且對客戶進行相關產品的推薦。本研究使用不同的分群方法進行推薦品質的比較，實驗結果證明使用本研究提出的方法，可有效提升客戶銷售價值與客戶忠誠度。

As the high tech market change quickly, companies are increasingly facing competitive pressures. Thus, it is important to find out the potential demands by past consumer sales data and react more promptly to satisfy customer demands. In this research, we apply data mining techniques to support sales strategies of analog IC design industry. We use K-means to cluster customers into different clusters based on customers’ RFM values. Then association rule mining is applied to discover buying association patterns from the sales data in each cluster. Based on the association patterns, recommendations are conducted to recommend products to customers. This thesis has compared different recommendation methods under different clustering effects. The experimental result shows that the proposed approach is promising to promote customer loyalty and sales.