Dynamics Community Forum Thread Details

Hello,

The forecasting functionality in application can be used to create sales or production forecasts, in combination or independently. For example, most make-to-order companies do not carry finished goods inventory, because each item is produced when it is ordered. Anticipating orders (sales forecasting) is critical for a reasonable turnaround time on the finished goods (production forecasting). As an example, component parts with lengthy delivery times, if not on order or on inventory, can delay production.

The sales forecast is the sales department's best guess at what will be sold in the future, and is specified by item and by period. However, the sales forecast is not always adequate for production.
The production forecast is the production planner's projection of how many end items and derived subassemblies to produce in specific periods to meet the forecasted sales.

In most cases, then, the production planner modifies the sales forecast to fit the conditions of production, yet still satisfies the sales forecast.

You create forecasts manually on the Demand Forecast page. Multiple forecasts can exist in the system, and are differentiated by name and type. Forecasts can be copied and edited as necessary. Note that only one forecast is valid for planning purposes at a time.

The forecast consists of a number of records each stating item number, forecast date, and forecasted quantity. The forecast of an item covers a period, which is defined by the forecast date and the forecast date of the next (later) forecast record. From a planning point of view, the forecasted quantity should be available at the start of the demand period.

You must designate a forecast as Sales Item, Component, or Both. The forecast type Sales Item is used for sales forecasting. The production forecast is created using the Component type. The forecast type Both is only used to give the planner an overview of both the sales forecast and the production forecast. With this option, the forecast entries are not editable. By designating these forecast types here, you can use the same worksheet to enter a sales forecast as you do a production forecast, and use the same sheet to view both forecasts simultaneously. Note that the system treats the different inputs (sales and production) differently when calculating planning, based on item, manufacturing, and production setup.

Below is the new Extension from Microsoft on Sales and Inventory Forecast in Dynamics 365 Business Central. The Sales and Inventory Forecast extension enhances inventory management. The extension uses Azure Artificial Intelligence to predict future sales based on sales history to avoid inventory shortage and maintain sufficient working capital. Using a job queue, the user can automatically update predictive forecasts. This blog will explain how to use the Sales and Inventory Forecast extension in Dynamics 365 Business Central.

Where do you find the Extension?

The Microsoft extension is typically pre-installed in “installed extensions”:

The extension is available in both the production and sandbox environments.

The extension considers previous transactional data pertaining to the item and makes predictions with variance ranges for future coming periods.

Sales and Inventory Forecast Parameters

On the Sales and Inventory Forecast Setup page, the user can control key parameters.

Setting Period Type, Horizon, and Stock Warning Horizon determine how far into the future the forecasts cover and when items would run out of inventory. In this instance, stockout warning occurs for up to 3 months out, and the total time covered would be 12 months out.

Under Variance %, the user can control for the deviance in error, plus or minus. Any variances over 40% would lead to the application not making a forecast at all, due to the variance being too high. In this case, the system on the item page, fact box, will generate an error:

The application will also not make a forecast if there is a lack of sufficient prior data. This data is based on item ledger entries, for sale type transactions with items involved.

Expiration Period can set number of days until the forecast disappears, as new forecasts would be made.

Historical Periods indicate how many past prior periods of data for which the application would gather details for future forecasts.

And finally, the Timeseries Model specifies the algorithm used for the time series analysis. More information is available for these models can be found in this Machine Learning Mastery article 11 Classical Time Series Forecasting Methods in Python (Cheat Sheet).

Set Up the Job Queue

Now with the Sales and Inventory Forecast Setup page entered with parameters, set up the job queue. In the Setup page, select new and then select Setup Scheduled Forecasting:

With status on hold, the user can set up parameters to run the sales forecast:

In the above setup, the forecast is run daily overnight, with 3 attempts before stopping should an error arise. When the setup is complete, change the status from on hold to ready via the process tab.

On item cards fast tabs, the forecast will yield a predictive graph:

From the forecast, the user can create purchase documents with quantities pre-filled based on missing quantity to prevent stockout inventory:

If the user creates a purchase invoice and purchase order but does not invoice it and leaves the document open and outstanding, the system will not account for those inventory quantities. Therefore, if the user were to choose to create a purchase invoice or purchase order before posting the prior entries for the item, the system would generate the same quantities for the item. This would also apply to sales documents in the same manner.

Hope this all helps.

Thanks,

Steve