The first session in our webinar series discussed the basics of setting up and analyzing your data to get an approximation on the “forecastability” of your planning combinations and further stratifying those combinations into ABC/XYZ classifications.

This second session goes deeper into assessing the available historical demand and  the rationale for how much history is sufficient in different scenarios.

Demand history, and how it is “bucketed,” makes a huge difference in both the robustness of your statistical model and its potential for delivering a “good” forecast ongoing. However, sometimes too much history is detrimental and 0’s are significant!

You will understand how forecasting in weekly or monthly buckets and how your supply agility can impact the setup of determining the amount of history is appropriate in any given situation. You will then be led through that detailed setup when pulling data from your planning system into Excel to do the analysis and sorting into historical horizon groupings. This second session sets the stage for statistical model assignment and execution that will be covered in the remaining two sessions in this series.

Some specifics that will be addressed in this webinar session include:

• Understand the role that historical horizon plays in determining appropriate statistical models
• Discover how differences in the way history is collected, and how it is provided to supply, can impact your “bucketing” decisions
• Learn how to perform detailed analyses of history in Excel to determine historical horizon selection
• Recognize that having  0’s in your history, and seeing when they occur, can impact statistical modeling assignment
• Get proven guidelines in structuring history and “tagging” combinations to improve statistical model selection
• Discuss appropriate frequency for analysis, statistical modeling runs and release of forecast to supply