Implementing Predictive Step in Multi Actions
The introduction of Predictive Scenario in SAP Analytics cloud last year made developers & consultants a soothsayer. It made SAP Analytics Cloud more fun to work with. On the other hand, Multi Actions helped planners to save time by running multiple data actions with various models & versions on the go. Previously, users were able to train and forecast the data manually in the Predictive Scenario tab. In the release of Q1 2022 of SAP Analytics Cloud, user can now trigger the prediction scenario right from the story itself with the help of Multi Actions. In this blog let’s see how to consume and use predictive step in multi action with an example of sales data.
Blogs to refer before jumping into the tutorial
In this blog we try to run prediction on sales data of a company and use that as a planning version by increasing it by a numeric factor. We try to achieve this process by just a single click in a story.
To start with we need a good dataset for prediction. I have taken a Sales dataset from Kaggle (See dataset). You can take any dataset but make sure you have quality data for running prediction. Import the dataset to SAC and create a model with planning option enabled.
Creating Predictive Scenario
First, we need to create a predictive scenario to consume it in multi action step. For that we will create a time series forecast based on the model. Time series forecast requires at least one date dimension in your model.
Step 1: Create a new time series predictive scenario by clicking Predictive Scenario from the side panel
Step 2: Now let’s add data source / model to the scenario
Step 3: After selecting the model, fill the required fields with proper data. Here let’s predict total sales for next one year.
Step 4: Just save the prediction scenario, do not train & forecast.
To get detailed context on Time series forecast scenario follow this blog
To see the result of predictive scenario we need a story. Let’s create a simple story with a table.
Step 1: Create new story and insert a table. Add necessary measures and dimensions. For our case add Total, Year & Version in columns. Add Product line in Rows.
Step 2: Let’s create a new version to save the predicted value. Click on tools and select version management and create a blank version then publish it.
Creating Data Action (Optional step)
Now let’s create a data action to increase the predicted sales by a factor.
Step 1: Create a new Data Action. Create a parameter for factor.
Step 2: Add an advanced formula step. (Target sales = Source sales * Factor)
Creating Multi Action
After creating predictive scenario, we need to consume it in multi action. Now, we have to create new multi action by click it on the side tab.
Step 1: Give a proper name and description for the multi action. Save it. Multi action gives privilege to add three types of steps.
Step 2: Click on “Add Predictive step” to create a step on predictive scenario
Step 3: Give desired name to the step. Then select the predictive scenario which we created few steps back. Select the model generated by scenario. Select a version to save the forecast (use the version created few steps back).
Step 4: We can now add the data action step to increase the predicted sales value by a factor. Select the data action from the drop down. Fill up the parameter value for data action.
Check the follow up action box to publish the target version automatically.
Save the multi action. Now the multi action is all set up to be executed in a story.
Inserting Multi Actions into the story
Step 1: Open the story which we created earlier in this blog. Go to insert -> select Planning Trigger -> Select Multi Action Trigger
Step 2: Give a name and description. Then select the multi action we created from the dropdown
Step 3: Let’s run the multi action to see the result
After successful run of multi action, we got the predicted total sales for 2022 published to Predicted version.
In this blog, we have explored a new feature in multi action to add predictive scenario as a step. This feature is available from Q1 2022 release. It makes the process of Prediction and Planning much easier and saves time.