長久以來，存貨管理都是備受關注的問題，在面對顧客需求不確定的情況下，存貨問題會影響整體成本，所以如何在考慮服務水準下設置庫存水準儼然成為一個重要的課題。
在本研究中，將考慮一個定期檢視政策下的庫存問題，假設管理者以基於一個關鍵分位數(critical fractile)的訂購目標水準(Order-up-to)做為我們的庫存策略，若市場上的需求衝擊影響為暫時的效應，潛在的需求趨勢會回歸至長期均衡的狀態，我們選擇使用穩定模型來設定有效庫存水準；反之，若需求衝擊影響為永久性的，需求趨勢會受到衝擊而產生持續性累積的效應，並隨著時間推移，需求趨勢會逐漸偏離平均值。那麼此種狀態下，我們就要選擇使用非穩定的模型來設定有效庫存水準。在此兩種不同狀態下的需求行為，對於庫存水準的設定也需有所不同，使其可因應前置期間的需求。其中每個時期的需求過程都是透過自我迴歸模型來建立，以自我迴歸模型中的參數ϕ為需求衝擊的影響，並使用貝氏架構隨著新的需求資料加入，更新自我迴歸中的參數值，以利於我們對模型的選擇，以及設定前置期間需求的存貨水準上也會更加準確，並可有效降低存貨以及缺貨成本。

Inventory management has long been a concern. In the face of demand uncertainty, inventory problems will affect the overall cost, so how to set the inventory level under consideration of a desired consumer service level has become an important issue.
We consider a periodic-review Inventory assume that the managers use an Order-up-to base stock policy based on a critical fractile as our inventory strategy. If the random demand shocks have a temporary effect, and the potential demand trend will return to the long-term equilibrium state, We choose to use the stationary inventory model to set the effective inventory Order-up-to level. Conversely, if the demand shock impact is permanent, the shock contain an element that represent a permanent departure from previous levels, and over time, the demand trend will gradually deviate from the average. Then in this state, we have to choose an non-stationary inventory model to set the effective inventory level. These two different states of demand need to be different for the inventory level , so that it can respond to the demand during the lead time.
The demand process in each period is established through the autoregressive model. The parameter ϕ in the autoregressive model is the impact of the demand shock, and the Bayesian structure is used to update the parameter values of the autoregressive model with the new data. The setting of inventory level will be more accurate and can effectively reduce the cost of inventory.

[1] Azoury, K.S., “Bayes solution to dynamic inventory models under unknown demand distribution.”, Management Science, 31, pp.1150–1160, 1985.
[2] Azoury, K.S., J., Miyaoka, “Optimal policies and approximations for a Bayesian linear regression inventory model.”, Management Science, 55, pp.813–826, 2009.
[3] Charnes, J.M., H., Marmorstein, W., Zinn, “Safety stock determination with serially correlated demand in a periodic review inventory system.”, European Journal of Operational Research, Soc. 46, pp.1006–1013, 1995.
[4] Dekimpe, M.G., D.M., Hanssens, “Empirical generalizations about market evolution and stationarity.”, Marketing Science, 14, pp.109–121, 1995a.
[5] Dekimpe, M.G., D.M., Hanssens, “The persistence of marketing effects on sales.”, Marketing Science, 14, pp.1–21, 1995b.
[6] Dekimpe, M.G., D.M., Hanssens, “Sustained spending and persistent response: a new look at long-term marketing profitability.”, Journal of Marketing Research, 36, pp.397–412, 1999.
[7] Guler, M.G., “A note on the effect of optimal advertising on the distributionfree newsboy problem.”, International Journal of Production Economics, 148, pp.90–92, 2014.
[8] Hanssens, D.M., “Order forecasts, retail sales, and the marketing mix for consumer durables.”, International Journal of Forecasting, 17, pp.327–346, 1998.
[9] Iglehart, D., “The Dynamic Inventory Problem with Unknown Demand Distribution.”, Management Science, 10, pp.429-440, 1964.
[10] Khouja, M., S.S., Robbins, “Linking advertising and quantity decisions in the single-period inventory model.”, International Journal of Production Economics, 86, pp.93–105, 2003.
[11] Lee, C.M., S.L., Hsu, “The effect of advertising on the distribution-free newsboy problem.”, International Journal of Production Economics, 129, pp.217–224, 2011.
[12] Lovejoy, W.S., “Myopic policies for some inventory models with uncertain demand distributions.”, Management Science, 36, pp.724–738, 1990.
[13] Lee, Y. S., “Management of a periodic-review inventory system using Bayesian Model Averaging when new marketing efforts are made.”, International Journal of Production Economics, 158. pp.278-289, 2014.
[14] Marques, A., D.P., Lacerda, L.F.R., Camargo, R., Teixeira, “Exploring the relationship between marketing and operations: neural network analysis of marketing decision impacts on delivery performance.”, International Journal of Production Economics, 153, pp.178–190, 2014.
[15] Ma, P., H., Wang, J., Shang, “Supply chain channel strategies with quality and marketing effort-dependent demand.”, International Journal of Production Economics, 144, pp.572–581, 2013.
[16] Raftery, A.E., D., Madigan, J.A., Hoeting, “Bayesian model averaging for linear regression models.”, Journal of the American Statistical Association, 92, pp.179–191, 1997.
[17] Rossi, P.E., G.M., Allenby, R., McCulloch, Bayesian Statistics and Marketing., Wiley Series in Probability and Statistics, Wiley, Hoboken, NJ, 2005.
[18] Scarf, H., “Bayes solutions of the statistical inventory problem.”, The Annals of Mathematical Statistics, 30, pp.490–508, 1959.
[19] Scarf, H., “Some Remarks on Baye’s Solution to the Inventory Problem.”, Naval Research Logistic Quarterly, No .7, pp.591-596, 1960.
[20] Tang, C.S., “A review of marketing-operations interface models: from coexistence to coordination and collaboration.”, International Journal of Production Economics, 125, pp.22–40, 2010.
[21] Urban, T.L., “A periodic-review model with serially-correlated, inventory level-dependent demand.”, International Journal of Production Economics, 95, pp.287–295, 2005.