Creating an SPSS Modeler batch job with multiple data sources

Example

These steps demonstrate how to create the connections and identify the tables.

1. Create a connection in your project.

   To run the SPSS Modeler flow, you start in your project and create a connection for each of the three databases your flow connects to.

   For this example, the database connections in the project are named dashdb_conn, db2_conn1, and db2_conn2.

2. Configure the Data Asset nodes in your SPSS Modeler flow to use the connections.

   Configure each node in your flow to reference one of the three connections you created (dashdb_conn, db2_conn1, and db2_conn2), and then specify a table for each node.

   Note: You can change the name of the connection at the time of the job run. The table names that you select in the flow are referenced when the job runs. You can't overwrite or change them.

3. Save the SPSS model to the Watson Machine Learning repository.

   It's helpful to provide the input and output schema when you save a model. It simplifies the process of identifying each input when you create the batch job in the Watson Studio user interface. Connections that are referenced in the Data Asset nodes of the SPSS Modeler flow must be provided in the node name field of the input schema. To find the node name, double-click the Data Asset import node in your flow to open its properties:

   

   Note:

   SPSS models that are saved without schemas are still supported for jobs, but you must enter node name fields manually and provide the data asset when you create the job.

   This code sample shows how to save the input schema when you save a model (Endpoint: POST /v4/models).

        {
              "name": "SPSS Drug Model",
              "label_column": "label",
              "type": "spss-modeler_18.1",
              "runtime": {
                      "href": "/v4/runtimes/spss-modeler_18.1"
               },
               "space": {
                      "href": "/v4/spaces/<space_id>"
               },
               "schemas": {
                         "input":  [ { "id": "dashdb_conn", "fields": [] },
                                           { "id": "db2_conn1 ", "fields": [] } ,
                                           { "id": "db2_conn2", "fields": [] } ],
                          "output": [{ "id": "db2_conn2 ","fields": [] }]
                }
        }
   

   Note: The number of fields in each of these connections doesn't matter. They’re not validated or used. What's important is the number of connections that are used.

4. Create the batch deployment for the SPSS model.

   For SPSS models, the creation process of the batch deployment job is the same. You can create the deployment with the model that was created in the previous step.

5. Create SPSS batch jobs.

   You can create a batch job from the Watson Studio user interface or by using the REST API. If the schema is saved with a model, the Watson Studio user interface makes it simple to accept input from the connections specified in the schema. Because you already created the data connections, you can select a connected data asset for each node name field that displays in the Watson Studio user interface as you define the job.

   The name of the connection that is created at the time of job submission can be different from the one used at the time of model creation. However, it must be assigned to the node name field.

Creating jobs for models when schema is not provided

If the schema isn't provided in the model metadata at the time the model is saved, you must enter the import node name manually. Further, you must select the data asset in the Watson Studio user interface for each connection. Connections that are referenced in the Data Asset import nodes of the SPSS Modeler flow must be provided in the node name field of the import/export data references.

         {
                     "deployment": {
                                    "href": "/v4/deployments/<deploymentID>"
                        },
                        "scoring": {
                                    "input_data_references": [
                                       {
                                          "id": "dashdb_conn",
                                          "name": "dashdb_conn",
                                          "type": "data_asset",
                                          "connection": {},
                                          "location": {
                                                      "href": "/v2/assets/<asset_id>?space_id=<space_id>"
                                             },
                                             "schema": {}
                                    },
                                    {
                                             "id": "db2_conn1 ",
                                             "name": "db2_conn1 ",
                                             "type": "data_asset",
                                             "connection": {},
                                             "location": {
                                                      "href": "/v2/assets/<asset_id>?space_id=<space_id>"
                                             },
                                             "schema": {}
                                    },
                                    {
                                             "id": "db2_conn2 ",
                                             "name": "db2_conn2",
                                             "type": "data_asset",
                                             "connection": {},
                                             "location": {
                                                      "href": "/v2/assets/<asset_id>?space_id=<space_id>"
                                             },
                                             "schema": {}
                                    }],
                                    "output_data_reference": {
                                             "id": "db2_conn2"
                                             "name": "db2_conn2",
                                             "type": "data_asset ",
                                             "connection": {},
                                             "location": {
                                                      "href": "/v2/assets/<asset_id>?space_id=<space_id>"
                                             },
                                             "schema": {}
                                    }
         }