Lab: AX 2012 Cortana intelligence services...

12
Lab: AX 2012 Cortana intelligence services management - Recommendations with Customer service Time: 45 minutes In this hands-on-lab, you will take a tour of the Cortana intelligence service management and see how to work with Azure Recommendations SaaS service in Dynamics AX 2012 R3. You will walk-thru an end-to-end scenario based on the Call-center service recommendations that this sample application is designed to support. The overview provided in this lab will demonstrate the capabilities of the Azure Recommendations SaaS service and from an end-user point of view is meant to provide an example of a completed scenario. Note: You can run this lab in the cloud one-box environment provided. Overview This lab aims to demonstrate the Azure Recommendations related capabilities available within the sample AX module – Cortana intelligence services management. The module has a Azure Recommendations menu section that we will use in this lab to create a model and finally train for generating recommendations using a sample Retail catalog dataset. With this we aim to both simplify both managing, integrating and consuming Recommendations service. Key concepts Installing and configuring pre-requisites Provisioning an Azure Recommendations API account Azure Recommendations in Dynamics AX 2012 R3 Next steps Lab 5: Creating a model and data model elements 1

Transcript of Lab: AX 2012 Cortana intelligence services...

Page 1: Lab: AX 2012 Cortana intelligence services …ax-cis-ws.azurewebsites.net/modules/AX2012-Recommender... · Web viewAzure Recommendations in Dynamics AX 2012 R3 Next steps Installing

Lab:  AX 2012 Cortana intelligence services management - Recommendations with Customer serviceTime: 45 minutes

In this hands-on-lab, you will take a tour of the Cortana intelligence service management and see how to work with Azure Recommendations SaaS service in Dynamics AX 2012 R3. You will walk-thru an end-to-end scenario based on the Call-center service recommendations that this sample application is designed to support. The overview provided in this lab will demonstrate the capabilities of the Azure Recommendations SaaS service and from an end-user point of view is meant to provide an example of a completed scenario.Note: You can run this lab in the cloud one-box environment provided.

Overview

This lab aims to demonstrate the Azure Recommendations related capabilities available within the sample AX module – Cortana intelligence services management. The module has a Azure Recommendations menu section that we will use in this lab to create a model and finally train for generating recommendations using a sample Retail catalog dataset. With this we aim to both simplify both managing, integrating and consuming Recommendations service.

Key concepts

Installing and configuring pre-requisites Provisioning an Azure Recommendations API account Azure Recommendations in Dynamics AX 2012 R3 Next steps

Lab 5: Creating a model and data model elements

1

Page 2: Lab: AX 2012 Cortana intelligence services …ax-cis-ws.azurewebsites.net/modules/AX2012-Recommender... · Web viewAzure Recommendations in Dynamics AX 2012 R3 Next steps Installing

Installing and configuring pre-requisites The AX 2012 R3 CU 10 instance on your VM is re-built with a few prerequisites. Entity Store hotfix Cortana Intelligence Suite AX project Demo dataset

The only other requirement is to create a subscription key for the Azure Recommendations APIThat will be used to generate a token to sign the service requests.

Provisioning an Azure Recommendations API account

Launch the AX 2012 R3 client and navigate to the pre-installed “Cortana intelligence services management” module.

Click on the “Getting started with recommendations” menu item. This launches the public Azure Recommendations landing page.

Lab 5: Creating a model and data model elements

2

Page 3: Lab: AX 2012 Cortana intelligence services …ax-cis-ws.azurewebsites.net/modules/AX2012-Recommender... · Web viewAzure Recommendations in Dynamics AX 2012 R3 Next steps Installing

Click on the “Sign up” button on the right hand side to launch the Recommendations app in the Microsoft Azure data marketplace.

Click “Sign up” on the 10,000 transactions/month option on the right. Follow through by providing your email account details.

Lab 5: Creating a model and data model elements

3

Page 4: Lab: AX 2012 Cortana intelligence services …ax-cis-ws.azurewebsites.net/modules/AX2012-Recommender... · Web viewAzure Recommendations in Dynamics AX 2012 R3 Next steps Installing

NOTE: At the end of this sign-up you will be provided an account key. Carefully note down the account key and the email used for signing up

Switch back to the Ax Client and launch the – Setup > Global parameters form

Enter your own email and account key in the Recommendations API settings section in the Global parameters form. Close the form to persist the values.

Azure Recommendations in Dynamics AX 2012 R3In general, to use Azure recommendations, you need to follow these steps outlined below.

Lab 5: Creating a model and data model elements

4

Create model Import catalog data Import usage dataBuild a

recommendation model

Consume recommendations

Page 5: Lab: AX 2012 Cortana intelligence services …ax-cis-ws.azurewebsites.net/modules/AX2012-Recommender... · Web viewAzure Recommendations in Dynamics AX 2012 R3 Next steps Installing

These can be accomplished in AX using Cortana intelligence services management capabilities.

1. Create models

Click on the – Azure Recommendations > Recommender models menu to launch the Recommender models form

NOTE: The Action pane outlines the list of functionality that can be leveraged to create models and its associated artifacts.

Click “New” action pane menu to launch the Create new recommender dialog. Enter a valid recommender model name, give a description and enter a name for the specific model entity. Click OK.

NOTE: o Model name must be 3-20 characters longo For Entity enter value - RetailCatalog

Lab 5: Creating a model and data model elements

5

Page 6: Lab: AX 2012 Cortana intelligence services …ax-cis-ws.azurewebsites.net/modules/AX2012-Recommender... · Web viewAzure Recommendations in Dynamics AX 2012 R3 Next steps Installing

NOTE: The Entity attribute is currently a loose association that indicates the category or the entity that a model is being created for. The Recommendations API does not have any such notion and this is strictly only an AX attribute for enabling multi-category recommendations.

If the model creation is successful, then a new model is shown in the models list page. If models already exist, then clicking Synchronize would sync all models available

for the specific subscription.

2. Import catalog data

After a model is created, catalog data – the collection of all items is to be imported. In our lab, we will use a Retail catalog data file based on the DEMO dataset

In the Recommender models form click Upload catalog action pane menu button to launch the catalog file upload dialog.

Lab 5: Creating a model and data model elements

6

Page 7: Lab: AX 2012 Cortana intelligence services …ax-cis-ws.azurewebsites.net/modules/AX2012-Recommender... · Web viewAzure Recommendations in Dynamics AX 2012 R3 Next steps Installing

Browse and select the catalog data file (file attached inplace below). Provide a description for the data file and click OK.

After the import completes, click on the View file upload history menu to verify that the upload status for Catalog file is Completed.

3. Import usage data

The next step in the recommendations process is to import usage file data. This is typically the sales transactions or user clicks that enables the models to be trained for predictions.

In the Recommender models form click Upload usage file action pane menu button to launch the usage file upload dialog.

Lab 5: Creating a model and data model elements

7

Page 8: Lab: AX 2012 Cortana intelligence services …ax-cis-ws.azurewebsites.net/modules/AX2012-Recommender... · Web viewAzure Recommendations in Dynamics AX 2012 R3 Next steps Installing

Browse and select the usage data file (import each of the attached files below one at a time). Provide a description for the data file and click OK.

After the import completes, click on the View file upload history menu to verify that the upload status for all Usage file’s is shown Completed.

4. Build a recommendation model

The primary step in Recommendations workflow is to train the models. This is done by triggering appropriate builds on the model for generating recommendations. The Azure Recommendations supports 3 types of builds – FBT (Frequently bought together) or I2I, Recommendations or U2I and then Rank build. For details on these builds and their parameters see – Azure Recommendations Quick Start Guide.

In AX 2012, the model builds can be triggered by using the Azure Recommendations > Manage recommendation builds menu or by clicking the View model builds action pane menu on the Recommender models form

Lab 5: Creating a model and data model elements

8

Page 9: Lab: AX 2012 Cortana intelligence services …ax-cis-ws.azurewebsites.net/modules/AX2012-Recommender... · Web viewAzure Recommendations in Dynamics AX 2012 R3 Next steps Installing

Click New to launch the model build dialog. Choose “Recommendations build” option for Model build type. Specify a description for the build and click OK.

NOTE: Once the build is triggered a build id is generated. The list page is then refreshed with the new build details.

Click Refresh status to refresh the build status. Wait for the build status to be Successful.NOTE: Recommendations build should take < 5 minutes to build. Also note that you can have utmost one build of a specific type in progress at any given time.

Select the newly created build and click Set active build.NOTE: This is done since a model can have many builds of each of the supported types and it is expected that user selects a specific build to be used in a recommendation scenario.

Lab 5: Creating a model and data model elements

9

Page 10: Lab: AX 2012 Cortana intelligence services …ax-cis-ws.azurewebsites.net/modules/AX2012-Recommender... · Web viewAzure Recommendations in Dynamics AX 2012 R3 Next steps Installing

5. Consume recommendations

Once the model is trained (i.e., successful builds are generated), it is ready to be consumed in recommendation scenarios. In AX, the Retail recommendations can be provided using the POS interface or in offline scenarios where Call center service representative takes offline orders via a catalog. We will use the Call center service form to demo the recommendation.

Import the below XPOs and then compile the private project – CortanaIntelligenceCockpit from your AX developer workspace.

Switch to company USRT Launch the Customer service form from Call center > Journals > Customer service menu

Click Enable recommendations action pane menu in the Customer service form. Enter customer name - 1001 and click Search.

Lab 5: Creating a model and data model elements

10

Page 11: Lab: AX 2012 Cortana intelligence services …ax-cis-ws.azurewebsites.net/modules/AX2012-Recommender... · Web viewAzure Recommendations in Dynamics AX 2012 R3 Next steps Installing

In the Search form, select and click “Select customer” action pane menu Click View Recommendations action pane menu

Conclusion

This lab demonstrates how it is possible to incorporate Azure Recommendations in your Dynamics AX sales process to cross-sell and/or upsell products.

Lab 5: Creating a model and data model elements

11