| dc.contributor | 資管系 | |
| dc.creator (作者) | 莊皓鈞 | |
| dc.date (日期) | 2018-10 | |
| dc.date.accessioned | 10-Jun-2026 11:54:09 (UTC+8) | - |
| dc.date.available | 10-Jun-2026 11:54:09 (UTC+8) | - |
| dc.date.issued (上傳時間) | 10-Jun-2026 11:54:09 (UTC+8) | - |
| dc.identifier.uri (URI) | https://ah.lib.nccu.edu.tw/item?item_id=182867 | - |
| dc.description.abstract (摘要) | 零售店內的作業除勞力密集，牽涉到員工、流程、科技，有不少環節容易出錯，即使是 Walmart 這般成功的實體零售商，店內營運從結帳掃描、貨架填補到後端收貨的執行錯誤相當常見，因錯誤發生的架上缺貨(shelf out-of-stock)更造成了顯著的獲利損失。針對架上缺貨這個困擾零售商、顧客與供應商的問題，本研究發展決策支援模型，供給欲透過資料驅動稽核來修正架上缺貨的零售業經理人參考。在文獻中常見的iid 間斷性需求結構下，我們使用POS 資料中的連續零銷售觀察值，發展出考量成本的稽核政策，此政策平衡了執行稽核的邊際成本與潛在架上缺貨持續的邊際成本。在推導出最佳解與探索政策性質後，我們解放常被使用的iid 假設，運用整數自我相關的隨機過程，發展出AR(1)間斷性需求結構下的稽核決策模型。本研究將探索模型中參數群對最佳稽核政策的直接與交互作用，並解析最佳解的性質。此外為了深入瞭解間斷性AR(1)一般化iid 的效益，我們預期將推導出成本差異指標，用其評估在不同的自我相關程度和需求分散度(dispersion)下，誤用iid 間斷性稽核政策所造成的損失。最後，本研究將用真實銷售資料估計整數自我相關需求模型，瞭解間斷性AR(1)模型的實務需求度。 | |
| dc.description.abstract (摘要) | Retail store operations are labor-extensive and error-prone as store operations involve people, processes, and technology. Execution errors such as misreporting sales and misplacing products have become norms even at successful retailers. Among documented symptoms of poor store operations, shelf out-of-stock (OOS) is a salient operational problem that causes non-trivial profit loss in retailing. To tackle shelf-OOS that plagues customers, retailers, and suppliers, we develop a decision support model for managers who aim to fix the recurring issue of shelf-OOS through data-driven audits. Under a commonly-assumed i.i.d. demand structure, we use consecutive zero sales observations in point-of-sale (POS) data to develop a cost-sensitive policy, which strikes the balance between marginal cost of auditing an item and marginal cost of not fixing probable shelf-OOS. After deriving the audit policy and showing its comparative statics, we relax the discrete i.i.d. demand assumption and develop a cost-sensitive audit policy in the autoregressive discrete demand scenario. Specifically, we model demand observation as a discrete time, non-negative, and integer-valued process with first-order autocorrelation (INAR(1)). We plan to derive comparative statics that depict policy behaviors and perform an extensive numeric study that reveals interactions between policy parameters. To assess the utility of the INAR(1) generalization of the i.i.d., we will devise cost differential metrics and formally quantify the economic benefits of the INAR(1) model under various levels of autocorrelation and dispersion in discrete demand. Finally, we aim to estimate INAR(1) models for various retail demand observations and provide empirical evidence for violation of the discrete i.i.d assumption, which strengthens the practical need for the proposed policy. | |
| dc.format.extent | 116 bytes | - |
| dc.format.mimetype | text/html | - |
| dc.relation (關聯) | 科技部, MOST106-2628-H004-002, 106.08-107.07 | |
| dc.subject (關鍵詞) | 零售業營運; 最佳化貨架稽核; 整數自我相關; 資料分析 | |
| dc.subject (關鍵詞) | Retail operations; Optimal shelf audits; Integer-valued autoregressive processes; Data analytics | |
| dc.title (題名) | 連續零銷售與零售貨架稽核 | |
| dc.title (題名) | Consecutive Zero Sales and Retail Shelf Audits | |
| dc.type (資料類型) | report | |
