Exporting JSON for a cloud-hosted Machine Learning endpoint

You export a JSON file for configuring a cloud-hosted Machine Learning endpoint.

Before you begin

You must download the model from IBM Cloud Pak® for Data External link opens a new window or tab. You must have an account with IBM Cloud Pak for Data to use the resource. Contact your IBM® representative for an account.

About this task

You use IBM Cloud Pak for Data to obtain the JSON file.

Procedure

To get the JSON file:

  1. Open IBM Cloud Pak for Data External link opens a new window or tab.
  2. From the Navigation Menu "" in the top left corner, choose Deployments.
  3. Select your deployment space.
  4. Choose your model in the Assets list.
  5. Click your deployment in the list.
  6. On the API reference page, the public and private endpoint of the deployment are displayed.
  7. Copy the value of the public endpoint field. You use this value later to make a separate note of this part of the URL: 
    https://us-south.ml.cloud.ibm.com/ml/v4/deployments/
30d8221b-47ad-48d1-8b68-2f042474ff91
predictions?version=2021-05-01
  8. Note down the deployment ID (for example, 30d8221b-47ad-48d1-8b68-2f042474ff91).
  9. Click your deployment space in the breadcrumbs at the top of the browser window: Deployments / <your deployment space> / Mortgage Application Prediction Model.
  10. Click the arrow button "" above the tray at the top right.
  11. Choose Export space.
  12. Click the New export file button.
  13. Give a name for your export file and select the check box next to Mortgage Approval Prediction Model.
  14. Click the Create button.
  15. From the entry that appears in the table, hover over it and click the Download button to the right of the entry.
  16. Open the download directory on your computer and locate the file with the name you chose. Open the compressed (.zip) file.
  17. Inside the compressed file, navigate to assets/.METADATA.
  18. Extract the file wml_model in the directory. It contains the details of your data model. It is not well formatted. To view it, you can use it any JSON formatter of your choice. It can be left unformatted if you do not want to inspect it.