Decision Model

This activity allows you to utilize an existing machine learning (ML) model designed using Albie Machine Learning Workbench that is deployed on the ML server such as Rapidminer server. It helps you to retrieve prediction on the new data provided.

Configuring the ML server and the ML model in automation Studio deploys them as a web service creating the REST API URL. The supported http request methods that uses the web services corresponding to the ML model are:

·       GET - predictions are made by passing input fields as the URL query parameters

·       POST - predictions are made by passing the JSON file

You can provide the set of data required for the ML model to make the prediction and return the response in the JSON format. The JSON activity can then be used to retrieve the required output fields from the ML model JSON response and decide to either complete the task or involve human inputs by using the Decision Workbench activity that ends the flow of the automated process and creates a task in the Albie Decision Workbench application. See Decision Workbench activity to know how to use the activity.

Prerequisite

Configure the ML server  in  Automation Studio to utilize the web services of the available ML model. Below are the minimum required properties for configuring the intended ML server. Remaining fields are auto populated with the default values already configured in Automation Studio . If you want to change the default values, refer Contextual Intelligence Apps in Admin Capabilities section.

1.    In the Admin menu, add an application of Application Type- Contextual Intelligence Apps. The mandatory fields are highlighted with red box.

2.    In the Application Properties panel:

·       In the HTTP Method list, select the request method. Available options are, GET and POST.

·       In the Server Url field, enter the URL of the server where the ML models are deployed.

·       Select the Signin checkbox  if sign in to the server is required.

·       In the Display Name, enter a desired name of the server.  

3.    Click the (Save Properties) icon to save the application details.

The ML server is configured in Automation Studio.

Using Decision Model Activity

1.    Within the required automation process workflow, click Process Components in the Canvas Tools pane to expand the tool and view the associated activities.

2.    Drag the Application activity and drop on to the Flowchart designer on the Canvas.

3.    In the Application Type list, select Contextual Intelligence Apps.

4.    In the Select An Application list, select the configured ML server.

5.    Double click the Application activity, drag the Decision Model activity and drop inside the Application activity.

6.    In the ML Server Username field, enter the username of the configured ML server .

7.    In the ML Server Password field, enter the password of the configured ML server.

8.    Click GET MODELS to populate the available ML models in the Model list.

9.    In the Model list, select the required ML model as per your business need.

10.  Click the (Model Input and Response) icon to configure the ML model inputs and the output. The ML Model Inputs and Response (GET) or the ML Model Inputs and Response (POST) dialog box appears, depending on the selected  request type.

The configuration of the ML model inputs and output varies as per the selected request method. Click any of the links below to know the step-by-step process related to the selected request type:

·       GET

·       POST

Decision Model Activity Properties

The properties of a Decision Model activity are listed in the following table and can be edited in the Properties grid on the right panel.

   Related Topics

Step-By-Step Guide to Use Decision Model and Decision Workbench to Predict Whether an Industrial Drilling Machine Requires Maintenance

Decision Workbench