A new approach to forecasting intermittent demand based on the mixed zero‐truncated Poisson model

• Author: Kwok Leung Tam,Yufeng Zhang,Xiaoyun Guo,Aiping Jiang
• Published date: 01 January 2020
• DOI: http://doi.org/10.1002/for.2614
• Date: 01 January 2020

RESEARCH ARTICLE

A new approach to forecasting intermittent demand based

on the mixed zero‐truncated Poisson model

Aiping Jiang

1

| Kwok Leung Tam

2

| Xiaoyun Guo

1

| Yufeng Zhang

1

1

SHU‐UTS SILC Business School,

Shanghai University, Shanghai, China

2

School of Mathematics and Statistics,

University of Sydney, Camperdown, New

South Wales, Australia

Correspondence

Aiping Jiang, SHU‐UTS SILC Business

School, Shanghai University, 20

Chengzhong Road, Jiading District,

Shanghai 201899, China.

Email: ap724@shu.edu.cn

Abstract

This paper proposes a new approach to forecasting intermittent demand by

considering the effects of external factors. We classify intermittent demand

data into two parts—zero value and nonzero value—and fit nonzero values

into a mixed zero‐truncated Poisson model. All the parameters in this model

are obtained by an EM algorithm, which regards external factors as indepen-

dent variables of a logistic regression model and log‐linear regression model.

We then calculate the probability of occurrence of zero value at each period

and predict demand occurrence by comparing it with critical value. When

demand occurs, we use the weighted average of the mixed zero‐truncated

Poisson model as predicted nonzero demands, which are combined with pre-

dicted demand occurrences to form the final forecasting demand series. Two

performance measures are developed to assess the forecasting methods. By pre-

senting a case study of electric power material from the State Grid Shanghai

Electric Power Company in China, we show that our approach provides greater

accuracy in forecasting than the Poisson model, the hurdle shifted Poisson

model, the hurdle Poisson model, and Croston's method.

KEYWORDS

demand forecasting, intermittent demand, fitting distribution method, hurdle model, zero‐truncated

model

1|INTRODUCTION

The features of intermittent demand are characterized by

their irregularity, with a very small demand size. In addi-

tion, intermittent demand is greatly affected by external

factors. Traditional prediction techniques, such as

Croston's method and its variations (Boylan & Babai,

2016; Cameron & Trivedi, 1998; Croston, 1972), are not

applicable in such circumstances because they forecast

demand using time series data without considering exog-

enous factors (Croston, 1972). Only some of the literature

on the prediction of intermittent demand, such as Hua

and Tan (2007) and Hua and Zhang (2006), aim to gener-

ate data using time sequences with affecting external

factors taken into account. However, in such literature,

data based on intermittent demand time series are broken

down and prediction is not determined directly using the

data themselves. So far, to the best of our knowledge,

there has been little research on an intermittent demand

forecasting method that utilizes time series data directly

and takes external factors into account.

In this paper, we propose a fitting distribution forecast-

ing approach aimed at generating data on the intermit-

tent demand time series. There are two contributions in

this paper. Firstly, by taking external factors into consid-

eration, the method precisely forecasts the time of

demand occurrence. By using the hurdle concept, we

classify the intermittent demand data into two parts—

Received: 23 October 2018 Revised: 28 April 2019 Accepted: 10 May 2019

DOI: 10.1002/for.2614

Journal of Forecasting. 2020;39:69–83. © 2019 John Wiley & Sons, Ltd.wileyonlinelibrary.com/journal/for 69