Basics of Building and Analyzing Adaptively Targeted Forecast Models for Supply Chain Management

Abstract

Abstract— Forecasting is an under estimated field of research in supply chain management. To describe the methodology for building adaptively targeted forecast models based on the recursive least squares method and to show the possibility of using these models in economic analysis. Two cases were studied, which include targeting by a single given target value and by a target trajectory described by several consecutive values. It was shown that for the autoregressive model in the case of setting several target values, the multistep procedure of the recursive least squares method is not applicable. It was also possible to clarify the necessity of introducing changes into the adaptive regression analysis scheme for the case when the adaptively targeted model is built on the basis of the autoregressive one. Procedures for building adaptively targeted models for supply chain management of setting target conditions have been proposed. The adaptive regression analysis technique has been modified for the case of an adaptively targeted autoregressive model.