Demand Forecasts using Data Science & Artificial Intelligence

Do you play with the idea of taking your demand planning to a new level with artificial intelligence to

• optimize the collaboration between production planners and algorithms,
• reduce storage and depreciation costs,
• ensure your ability to deliver
• and save time on manual demand planning efforts?

On this page, we will inform you about what Demand Forecasting is all about. We will also introduce the relevant terminology, provide methodical insights, and more.

Content

• → A typical 'demand planning question' and solution
• → Blog articles on 'demand forecasting'
• → Video material and additional information
• → Experiences of our customers

A typical “demand planning question” and solution

Problem
A company produces a large number of items of different types. This includes articles with regular and seasonal demand as well as articles with sporadic demand. The portfolio contains both new items and long-standing blockbusters, trendy and stable items. What is the expected quantity for each article group in the next few weeks?

Solution
The ideal solution would be a (partially) automated system producing forecasts at article group level for the next weeks and thus doing justice to the different types of planning objects. Ideally, methods from statistics as well as machine learning and pattern recognition methods are used to meet the special challenge of having both sporadic demand and regular demand patterns in the data. In addition, existing open orders are included in the forecast. A classification mechanism determines which article groups can be predicted by artificial intelligence and how well, and for which additional validation by experts make sense. An analysis at different hierarchical levels determines whether the requirements of the various article groups can be better predicted at the level of interest or, instead, top-down or bottom-up. The data is taken from the company’s ERP system, and the results of the sales planning are fed back there. The company uses its own reporting tools for the display and visualization of the results. A dashboard visualizes interim results for planners and points them specifically to those article groups for which it is worth taking an expert look at the forecasts again and adjusting them if necessary.

Blog articles on “demand forecasting”

The following blog articles are provided in German:

• Forecast horizon, granularity & Co.: technical terms simply explained - General terms on the subject of “forecasting”.
• How do you measure forecast quality? - Backtesting, MSE, MAPE and more, including the topic “sporadic demand”, which is often encountered in the context of demand planning.
• Intermittent demand: how rare events can be predicted with the right methodology - On the special methodological approach to forecasting intermittent time series.
• Artificial intelligence in the context of forecasting - learning for the future with machine learning.
• 10 characteristics of a good forecast - which forecasts demand planners can trust.
• Optimize safety stocks and thus also optimize warehousing with the help of forecasts.
• Make a lot out of little: the importance of small data - Even with a short data history, high-quality demand forecasts can be created.
• Dealing with Weekly Forecasts - How methods from statistics and AI deal with the forecast on a weekly basis.
• Training, testing, crosswise – By using cross-validation, the model and its parameters can be optimized.

Video material and additional information

• You can download a detailed case study on ‘Demand Forecasting and Supply Chain’ here for free in German: Case Study Supply Chain (PDF-Download)
• Krebs, K., & Christ, T. (2022). Der wahre Erfolgsfaktor für Sales & Demand Predictions heißt Augmented Intelligence. Wirtschaftsinformatik & Management, 14(3), 176-184. https://doi.org/10.1365/s35764-022-00406-6
• Blackburn, R., Lurz, K., Priese, B., Göb, R., & Darkow, I. L. (2015). A predictive analytics approach for demand forecasting in the process industry. International Transactions in Operational Research, 22(3), 407-428. https://doi.org/10.1111/itor.12122
• The following video conveys a deep dive into the topic supply chain planning with artificial intelligence with an excerpt from our supply chain planning solution: ‘Automated supply chain planning with the help of AI’:
  
  

Experiences of our customers