What are the benefits of using a hierarchical approach to forecast error decomposition?

Forecast error decomposition is a technique that helps you understand the sources and magnitude of errors in your demand forecasts. It can help you improve your forecasting accuracy, adjust your inventory levels, and optimize your supply chain planning. In this article, you will learn what are the benefits of using a hierarchical approach to forecast error decomposition, and how to apply it to your data.

1 What is forecast error decomposition?

Forecast error decomposition is a method of breaking down the total forecast error into different components, such as bias, trend, seasonality, and random noise. By doing this, you can identify which factors are causing the most error, and how they vary across different time periods, products, or regions. You can also compare the performance of different forecasting methods, and choose the one that minimizes the error for your specific situation.

Add your perspective

2 What is a hierarchical approach?

A hierarchical approach is a way of organizing your data into a hierarchy of levels, such as product categories, subcategories, and items, or geographic regions, countries, and cities. Each level represents a different aggregation or disaggregation of the data, and has its own forecast and error. A hierarchical approach allows you to decompose the forecast error at each level, and also to reconcile the forecasts across levels, so that they are consistent and coherent.

Add your perspective

James Hulett

Operations Professional specializing in planning, project management, and process improvement
(edited)
Report contribution
It is important to note that error tends to compound as you disaggregate. High level analytics may mask error that may be revealed at lower levels of aggregation. It is important to consider multiple levels of aggregation when making important forecasting decisions from the top of the hierarchy to the bottom.

Like

3 What are the benefits of a hierarchical approach?

A hierarchical approach to forecast error decomposition has several benefits. It can capture different patterns and sources of error at different levels of granularity, reducing the overall forecast error by combining information from different levels. This approach can also improve accuracy and reliability of forecasts by using a reconciliation method, such as optimal combination or minimum revision, to adjust forecasts at each level and minimize the error.

Add your perspective

Keith Holdsworth

Senior Supply Chain consultant specialised in delivering improvements in S&OP | IBP | Demand Planning | Supply Planning | Logistics | Training & Development | BCP | FMCG | Pharma | Chemicals
Report contribution
An additional benefit to a hierarchical approach is when you can match the hierarchy to decision making requirements. Further out, you may only need to know how much of a raw material (on a long lead-time) you need to buy - so converting your forecast into how much of that material is used in a product is a helpful factor. Closer in, you may have to decide which recipes you will use in what proportion - so you can order other ingredients you require that are on a shorter lead-time (and this recipe level may be helpful to plan formulation capacity). Finally, you need the forecast at a SKU level - to understand what pack size and label to use as you package the formulated product. The closer you get, the more detail you need...

Like

4 How to use a hierarchical approach?

To use a hierarchical approach to forecast error decomposition, you need to first define your hierarchy of levels and collect historical data for each level. Then, select a forecasting method for each level and generate forecasts and errors. After that, decompose the forecast errors at each level into components such as bias, trend, seasonality, and random noise using an additive or multiplicative decomposition. Subsequently, compare the forecast errors and their components across levels to identify the main sources and drivers of error. Finally, reconcile your forecasts across levels with an optimal combination or minimum revision to ensure that they add up correctly and minimize the error. Once completed, evaluate the performance of your reconciled forecasts and errors to improve your forecasting process and decisions.

Add your perspective

5 What are some challenges and limitations?

Implementing a hierarchical approach to forecast error decomposition can be complex and time-consuming, particularly if you have a large number of levels, products, or regions, or if you use different forecasting methods for different levels. Additionally, it can introduce uncertainty and variability in your forecasts and errors, depending on the choice of hierarchy, forecasting method, decomposition method, and reconciliation method. Furthermore, it can be challenging to interpret and communicate the results of the forecast error decomposition and reconciliation if there are conflicting or contradictory findings across levels or components.

Add your perspective

James Hulett

Operations Professional specializing in planning, project management, and process improvement
(edited)
Report contribution
This is a very important note. The cost of increasing the sophistication of the forecast hierarchy may outweigh the benefits. Depending on an organization's forecasting tools, processes, or methodology, complicating the forecast hierarchy may bind you needlessly complex practices. Make sure the juice is worth the squeeze before chasing fancy hierarchy models. Additionally, it its possible that various ways of introducing hierarchy by stratification, sales channels, or other criteria may obscure statistical relationships between products. For example, if a forecaster is highly primarily focused on a low level hierarchy, they may miss big-picture insights that come with taking a step back and examining demand at a higher level.

Like

6 Hereâ€™s what else to consider

This is a space to share examples, stories, or insights that donâ€™t fit into any of the previous sections. What else would you like to add?

Add your perspective

Directory

What are the benefits of using a hierarchical approach to forecast error decomposition?

1

2

3

4

5

6

1 What is forecast error decomposition?

2 What is a hierarchical approach?

3 What are the benefits of a hierarchical approach?

4 How to use a hierarchical approach?

5 What are some challenges and limitations?

6 Hereâ€™s what else to consider

Demand Forecasting

Rate this article

Thanks for your feedback

More articles on Demand Forecasting

More relevant reading