To aggregate or not to aggregate: Forecasting of finite autocorrelated demand

by   Bahman Rostami-Tabar, et al.

Temporal aggregation is an intuitively appealing approach to deal with demand uncertainty. There are two types of temporal aggregation: non-overlapping and overlapping. Most of the supply chain forecasting literature has focused so far on the former and there is no research that analyses the latter for auto-correlated demands. In addition, most of the analytical research to-date assumes infinite demand series' lengths whereas, in practice, forecasting is based on finite demand histories. The length of the demand history is an important determinant of the comparative performance of the two approaches but has not been given sufficient attention in the literature. In this paper we examine the effectiveness of temporal aggregation for forecasting finite auto-correlated demand. We do so by means of an analytical study of the forecast accuracy of aggregation and non-aggregation approaches based on mean squared error. We complement this with a numerical analysis to explore the impact of demand parameters and the length of the series on (comparative) performance. We also conduct an empirical evaluation to validate the analytical results using monthly time series of the M4-competition dataset. We find the degree of auto-correlation, the forecast horizon and the length of the series to be important determinants of forecast accuracy. We discuss the merits of each approach and highlight their implications for real world practices.



There are no comments yet.


page 16

page 18

page 36

page 37

page 38


Reframing demand forecasting: a two-fold approach for lumpy and intermittent demand

Demand forecasting is a crucial component of demand management. While sh...

Demand forecasting in supply chain: The impact of demand volatility in the presence of promotion

The demand for a particular product or service is typically associated w...

Effects of Aggregation Methodology on Uncertain Spatiotemporal Data

Large spatiotemporal demand datasets can prove intractable for location ...

Taxi demand forecasting: A HEDGE based tessellation strategy for improved accuracy

A key problem in location-based modeling and forecasting lies in identif...

Online Hierarchical Forecasting for Power Consumption Data

We study the forecasting of the power consumptions of a population of ho...

Ensemble Machine Learning Methods for Modeling COVID19 Deaths

Using a hybrid of machine learning and epidemiological approaches, we pr...
This week in AI

Get the week's most popular data science and artificial intelligence research sent straight to your inbox every Saturday.