Title: A responsive multiplicative Holt-Winters approach for enhanced forecasting accuracy

Authors: H.M. Emrul Kays; A.N. Mustafizul Karim; Mohd Radzi C. Daud

Addresses: College of Business Administration, International University of Business, Agriculture and Technology, Dhaka, Bangladesh ' Faculty of Science and Engineering, Queensland University of Technology, Brisbane, Australia ' Department of Manufacturing and Materials Engineering, International Islamic University Malaysia, Kuala Lumpur, Malaysia

Abstract: In today's competitive business world, the quest for a more reliable forecasting method is an ongoing process. As a classical and reliable technique the Conventional Multiplicative Holt-Winters (CMHW) approach has been widely used in which three different recursion processes are incorporated to estimate the level, growth rate and the seasonal parameter. A set of estimated initial values having the number of elements equivalent to the seasonal length is used to initiate recursion for seasonal parameter. But the recursion processes of level and growth rate, initiated by using the individual single stationary values, might reflect a strong adherence to the preceding patterns of data sets compromising the forecasting accuracy. In this context the CMHW approach is modified with a periodic updating procedure in recursion process of the level and growth rate values. The proposed Responsive Multiplicative Holt-Winters (RMHW) method with a dynamic adjustment feature is validated through various sets of data.

Keywords: seasonal patterns; demand forecasting; smoothing constants; multiplicative Holt-Winters method; forecasting accuracy.

DOI: 10.1504/IJISE.2020.106850

International Journal of Industrial and Systems Engineering, 2020 Vol.35 No.1, pp.114 - 134

Received: 28 Nov 2017
Accepted: 25 Aug 2018
Published online: 23 Apr 2020 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article