近年來,無店舖零售業的興起(如:網路購物、電視購物)已形成一股重要的趨勢。由於消費者透過無店舖通路購物,無法親自檢視實際產品,使得消費者風險大為提高,導致無店舖通路業者為了有效與實體店舖競爭,降低消費者風險,博取消費者對產品的信任感,多會訂定7-10天內可無條件退貨等寬鬆的退貨政策,然而,這樣的政策卻也連帶著使退貨率不斷提高。對無店舖零售業而言,節節攀升的退貨率,加上逐漸高漲的退貨成本和遞減的邊際利潤,產品退貨的有效管理已變成無法忽略的議題。 因此,本研究以無店舖零售業的「退貨」為研究主題,採用羅吉斯迴歸與決策樹兩種方法,建構能有效預測消費者退貨傾向之模型,並以實證資料來驗證模型之判別能力預測準確度。研究結果顯示,羅吉斯迴歸分析與決策樹方法皆具有一定程度的預測能力,且兩者之預測績效並無差異,均可有效協助無店舖零售業建立退貨率預測模型,並作為解決高退貨率問題時的參考依據。
With the advent of information and communication technology, nonstore retailing has gained significant growth in recent years. While enjoying the convenience of distant purchasing, consumers however can not see and evaluate goods purchased prior to making the buying decision. The higher consumer’s risk has prompted most nonstore retailers to adopt somewhat lenient merchandise return policies (e.g., 7-10 days, no-question-asked return policy). However, this has resulted in an even higher return rate and hence higher operating costs and thinner profit margin. How to effectively manage customer’s intentional return of merchandise has become one of the major issues any nonstore retailer needs to consider. The main thrust of this research is to develop effective prediction model for customer’s return propensity. Here, two prediction models, based on logistic regression and decision tree, are proposed. A set of real data collected from a well-known nonstore retailer in Taiwan is used to validate the applicability of the proposed models. Our empirical findings show that both models performs equally well, compared to that of a myopia decision.