Nonlinear identification of judgmental forecasts effects at SKU level

Abstract

Prediction of demand is a key component within supply chain management. Improved accuracy in forecasts affects directly all levels of the supply chain, reducing stock costs and increasing customer satisfaction. In many application areas, demand prediction relies on statistical software which provides an initial forecast subsequently modified by the expert's judgment. This paper outlines a new methodology based on State Dependent Parameter (SDP) estimation techniques to identify the non-linear behaviour of such managerial adjustments. This non-parametric SDP estimate is used as a guideline to propose a non-linear model that corrects the bias introduced by the managerial adjustments. One-step-ahead forecasts of SKU sales sampled monthly from a manufacturing company are utilized to test the proposed methodology. The results indicate that adjustments introduce a non-linear pattern undermining accuracy. This understanding can be used to enhance the design of the Forecasting Support System in order to help forecasters towards more efficient judgmental adjustments.

On the identification of sales forecasting models in the presence of promotions

“…This offers an informative and good explanation of the good performance of negative adjustments on forecasting accuracy. Previous references found that experts' negative adjustments were more accurate (Fildes et al, 2009;Trapero et al, 2011); here, we provide a feasible explanation why this is so, at least regarding adjustments due to promotional activity. Figure 4 depicts the overall results of the forecasting methods under study considering all the observations irrespective of whether they were promoted or not.…”

Do Experts' SKU Forecasts Improve after Feedback?

“…Examples of such instructions range from making lesser adjustments (Fildes and Goodwin, 2007;Franses and Legerstee, 2009), to making smaller-sized adjustments (Franses and Legerstee, 2010) or, in contrast, making larger-sized adjustments (Fildes et al, 2009;Trapero et al, 2010) and making lesser upward adjustments (Franses and Legerstee, 2009;Fildes et al, 2009). We believe that these instructions are all rather vague, hard to measure and to quantify and sometimes even contradictory.…”

Optimal ordering policy for newsvendor models with bidirectional changes in demand using expert judgment

“…Subsequently, these models are extended by imposing constraints for practical considerations. The constraints do not allow the revised order quantity to exceed an upper limit in the case of positive demand adjustment due to limited fund/storage capacity and to fall under a lower limit in the case of negative demand adjustment in order to maintain a target service level (Jammernegg and Kischka [26], Shi et al [27], Abdel-Malek and Otegbeye [28]). A one-dimensional search algorithm is proposed to find the optimal values of the Lagrangian multipliers required in the constrained optimization problems.…”