Parameter optimisation for a demand forecasting model for sales campaign products with the option of an additional purchase

Abstract

Companies launching special price offers for finished products during a limited time period are confronted with the problem of determining the purchase amount. This is especially the case if the special offer concerns a product, which does not belong to the regular range of products. The problem of determining the size of a purchase for a single period is known in the literature as the newsvendor problem. The company has the option to order more of the same or of a similar product if the demand seems to exceed the demand as forecasted before launching the offer. The earlier the signal of 'exceeding forecasted demand' is detected, the earlier the company can make an additional of alternative purchase. which reduces the probability of an out-of-stock event. A good forecasting system is required to support this type of decision-making. It is required that after only a few days a decision needs to be made whether to purchase extra stock and, if yes, also its size. Simulations are made using real-life demand patterns in combination with an optimisation technique in order to determine when the additional forecast is best made and the optimal values of the parameters. The objective function is related to the newsvendor cost function.