Contributors: Ruvisha Pillay, Bill Tonetti

This article aims to give you an understanding of the predefined attributes that are configured in all Pivot Forecasting models.

Attributes can be standard or unique. Standard attributes are included in all Pivot Forecasting models, while unique attributes are specific to a company’s data and implementation.

The three key attributes in all Pivot Forecasting models are:

Items in your Pivot Forecasting model will include the same items that are being planned in your inventory app.

Locations are either the base level locations from your inventory app, or a location grouping level. Consolidated national or regional locations, for example, may be more applicable to the forecasting processes.

Customer groups are defined within the model configuration. Typically a subset of “Key Customers” need individual planning, and are assigned to customer groupings during setup. The default grouping of "OTHER" is automatically assigned to any customers that are not assigned as being key.

Non-key attributes are descriptive. Examples of these are:

They “describe” the key attributes and any grouping dimensions that may have been brought in for planning. This data is sourced from your inventory app.

Non-key attributes include classifications for items. Examples of classifications are:

The ABC classification, HML velocity and stock status are sourced from your inventory app. Forecast Results and Forecastability are classifications generated as part of the Pivot Forecasting calculation process.

Forecast Results are the forecasting methods that are used. These can be found in the Forecast Results field of the Demand - Statistics Panel.

Usually, the Forecast Type, which is the requested statistical method, is “Expert”. The Forecast Results will display the method that the engine selected.

If the expert forecast type is overridden, the method that was chosen will instead reflect in the relevant fields. The statistical engine will generally apply the selected method, unless the history is not suitable for it to apply the selected method. In that case, the engine will use an alternative method. For example, if you select a particular method like “Seasonal”, the system will try to use that method but it may need to revert to a different method like “Level”, “Level-Trend”, or “None” if there isn’t enough history or the history isn’t seasonal.

Forecastability is based on the item’s R-Squared, which is a measure of the ratio of dispersion (series standard deviation) to the average (series mean). Forecastability can either be “Low,” “Medium,” or “High.”

Since all items and customers are sourced from your inventory planning app, the function for making copies of existing items and making new ones has been disabled. All items in the Pivot Forecasting model will have the Created Items designation of “NO.”

The Import Date shows the last time an item was imported from the data source, which is your inventory app. Items, locations or customer groups are deleted from Pivot Forecasting if they are no longer being sent from your inventory app with each import update. However, items will remain in Pivot Forecasting as long as there are historic or future Final Forecasts. Items such as this are designated as “orphaned.” A power user can remove all previous and future final forecasts if they want the item(s) removed from the model.

Pivot Forecasting is configured with two attributes that are intended to be edited by users within the app. These are called Z_Ad Hoc and Z_Follow Up Flag.

The easiest way to update editable attributes is in the Data – Attributes panel. Simply update the value for an item or a group of items by inputting the values in this panel. When you “Save”, your changes will update all of the items that are beneath the currently selected “Node” on the navigation “Tree.”

Have you read the related articles? Check out our Pivot Forecasting collection!