Browse Prior Art Database

System and method for augmenting demand forecast accuracy Disclosure Number: IPCOM000235916D
Publication Date: 2014-Mar-28
Document File: 3 page(s) / 77K

Publishing Venue

The Prior Art Database


Disclosed are a system and method to integrate of out-of-shelf (OOS) occurrence into the demand forecast model. The system and method correct the registered demand before it is used as the input of the demand forecast algorithm.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 52% of the total text.

Page 01 of 3

System and method for augmenting demand forecast accuracy

Demand forecast is a crucial and vital aspect of the retail industry . Currently, no statistical model is capable of precisely predicting future demand . The most used method applied is Exponential Smoothing, followed by Moving Average. Those methods are rudimentary approximations and provoke out -of-stock conditions and/or overestimated stock quantities due to the inherent errors in the demand estimation . In particular, demand forecast is imprecise because retailers do not account for variations taking place due to out-of-shelf events (i.e. due to sales losses that are not registered and unexpected sales that occur because some other similar product was not present on the shelf).

When faced with product unavailability, customers present the following recurrent behavior:

• Try to buy the same product at another store • Substitute the product for another one from a different brand • Substitute the product for another one from the same brand • Delay the purchase

• Do not purchase any product

This pattern of behavior clearly alters the registered demand not only during the periods

when the OOS is happening, but also immediately after the OOS event. This fact leads to imprecise demand estimation, since a forced condition limited the demand in a given period.

Existing solutions try to limit or calculate the associated error of the process of demand forecast, but do not address the demand fluctuations that cause the imprecision on the demand forecast. Other approaches fail to consider both external and internal factors of the supply chain that affect the demand curve . Another approach mainly verifies if there is a substantial discrepancy between the demand forecast of a group of products and the actual demand of a specific product . One idea presents a method based on a time series model and a K-Nearest Neighbor model in order to simulate the market trend of a certain product in the past and make a demand forecast for the future . Finally, one approach estimates demand by taking into account periods of out -of-shelf (OOS); that is, the amount of time for which shelves remain empty . In essence, the method considers OOS periods as being moments of time for which data about sales is missing. In order to estimate the absent values, the authors employ interpolation techniques. This solution disregard...