An Introduction to Demand Forecasting in Microsoft Dynamics AX (Part 1)
Demand forecasting can be immensely valuable to a business organization. It supports the overall Sales and operations planning process for a business and can directly impact and help the downstream business operations. At the very fundamental level, demand forecasting can tell a business what they need, when they need it and what they anticipate/predict delivering into the market. And even beyond that, it touches other functional areas of a business like human resources, engineering, supply chain, distribution and financials for example. Therefore, businesses require tools to analyze information and anticipate changes that are necessary to make better decisions to support future demand. With Dynamics AX 2012 demand forecasting and master planning tool, businesses can take their sales and operations planning process to a higher level of sophistication. Similar functionality is also available in the NEW Dynamics AX.
Having said that, in this first part of blog series, we will visit the world of Dynamics AX Demand Forecasting.
AX has always had the capability to capture and maintain demand forecast, however its generation has been left up to the users’ imagination. Let’s talk about general forecast before we delve deep into demand forecasting. In a nutshell, we will have the actual demand in the system accompanied by the firmed supply plus our demand forecast, which all will go through Master Planning to develop our planned supply. In this part of the blog, we will see how demand forecasting is generated out of AX, and the required set up and configuration before we can do that.
The entire process of how demand forecasting works in Dynamics AX follows several steps. And they are as following:
We will create a forecast model where we can send our forecast results
We need to separate our products into different item allocation keys
Then we have to set up the demand forecasting parameters in AX
Once we are done with configuring the parameter set up, a statistical baseline forecast can be generated
User can then make changes/modifications to the forecast in Excel and publish it to the OLAP cube
Finally, AX can import the updated forecast numbers back into Dynamics AX and review forecast accuracy through KPIs
Now that we have laid out the six steps to utilize demand forecasting in Dynamics AX, in this first part of the blog series, let’s go over the first three steps (setup and configuration) before running the demand forecasting in AX.
First and foremost, we need to set up a Forecast Model in AX where the forecast results can be sent. Start with the Inventory management area page > Setup group > Forecast folder > Forecast models menu item. Once we click on the Forecast models menu item, we will get the following Forecast models form, where a new forecast model can be created.
Item Allocation Keys
Once we are done with creating Forecast model, our second task would be to create the item allocation keys. Follow to the Inventory management area page > Setup group > Forecast folder > Item allocation keys menu item. Once we click on the Item allocation keys menu item, the following form is displayed.
Here our products can be separated into different item allocation keys. We can choose an item allocation key and then click the Assign items button in the menu bar, to move over items from available items to specific item allocation keys.
Demand forecasting parameters in Dynamics AX
At this point we can start doing setup and configuration of parameters in the Master planning module as necessary preparation for generating the statistical baseline forecast. Proceed to the Master planning area page > Setup group > Demand Forecasting folder > Demand forecasting parameters menu link.
This form is divided into three sections: General, Forecast dimensions and Item allocation keys, which we are going to look into more details.
These parameters set up the exact process of how we want the demand forecasting to work in AX 2012. The basic unit of the forecast and the demand forecast/demand forecast accuracy cubes can be identified.
Demand forecast unit
It’s very important to define the demand forecast unit correctly, since demand forecasting generates the entire forecast based on quantity itself only. It has been set to “each” in our example in the figure below.
The file path of the spreadsheet where the demand forecasting will get generated is also specified in this form as well.
Historical transaction types
Since demand forecasting can be used to forecast both dependent and independent demand, we can also choose the type of transactions we would like to include into the forecast. For example, a company in the bedding industry might want to include only the sales order history to generate their statistical baseline forecast. In that case, they could just check the Sales order box and leave the rest blank. That would be an example of independent demand.
At the same time, a company in the metals and mining industry might consider if critical sub-components can be added to item allocation keys and included in demand forecasting. Surely it can be done and in that case production line checkbox is checked and we will leave the rest blank. That would be an example of dependent forecast. As you can see, different transaction types can be chosen based on different industries/business processes and different preferences.
Forecast algorithm parameters
For the forecast generation strategy, we can just specify the basic copy of the forecast from last year or use the statistical forecast. Additionally, the algorithm can be specified. In our example below, the ARIMA algorithm is utilized.
Next, in the forecast dimensions parameter, the dimensions to be used in the cubes can be assigned from the available forecast dimensions. The dimensions can be added or changed depending on how we want the entire process to work. However, in case of any modifications we will have to rerun the demand forecast in order for those changes to take effect and it will copy over any changes we have made in the past.
Item allocation keys
Lastly, let’s consider the Item allocation keys options. Here the items can be included in the item allocation keys for forecasting purpose. Just like we saw in the General section, we can include certain historical transaction types to generate the statistical baseline forecast and also specify the forecast generation strategy and algorithms for different types of item allocation keys.
In the next part of the blog series, we are going to actually see how the rest of the demand forecasting process works in Dynamics AX.