Defining a BYOM Model - Teradata Vantage

ClearScape Analytics ModelOps User Guide

Deployment
VantageCloud
VantageCore
Edition
Enterprise
IntelliFlex
VMware
Product
Teradata Vantage
Release Number
7.0
Published
April 2023
Language
English (United States)
Last Update
2023-04-19
dita:mapPath
rfi1654194187578.ditamap
dita:ditavalPath
ayr1485454803741.ditaval
dita:id
B700-1175
  1. Select Projects in the Navigation bar.
  2. Click a project in the List of Projects.

    The List of Models for the selected project displays in the Work area.

  3. Select Define BYOM Model.
  4. In the Basic tab, set the properties:
    Property Description
    Name Specifies the BYOM model name.
    Description Specifies the description of the BYOM model.
    Format Lets you select the format of the BYOM model: PMML, ONNX, H2O, H2O_DAI, Python, R, SAS
    Based on the different format selected for this property, there are different properties need to be set in the Import sheet. See steps 6 to 8, respectively.
    Link to BYOM An optional property that allows linking to BYOM model workspace. It can be a GIT or DS workspace.
  5. Select Save Model & Import Versions.
  6. Set the properties in the Import model version window.
    Based on the selection of Format, different properties are needed.
    For SAS Import sheet, set the following properties
    Tab Property Description
    Details Database  
    Table Select table from the dropdown list of available tables inside the database.
    Models inside table  
    Model Monitoring Enable Model Monitoring Select the checkbox to enable.
    Model Type Select the type of the model: classification or regression.
    Training Statistics Select training datasets from the dropdown list for validating statistics.
    Prediction Expression Free text to input column expression for validating statistics.
    Model Evaluation and monitoring Select the checkbox to enable feature and prediction drift monitoring.

    This is used for computing statistics after importing model.

    Select to enable model evaluation and performance monitoring.

    If enabled, model will evaluate; otherwise, you can directly approve and then deploy model.

    Metrics Type Select the Metrics Type: default or custom.

    If custom is selected, then user has to upload evaluation files.

    5.import-sas_details
    5.import-sas_model-monitoring
    For PMML,ONNX, H2O, H2O_DAI Import sheet, set the following properties
    Tab Property Description
    Details External Id Specifies the model name.
    Tags  
    Upload model file Upload control to upload only model file.
    Model Monitoring Enable Model Monitoring Select the checkbox to enable.
    Model Type Select the type of the model: classification or regression.
    Training Statistics Select training datasets from the dropdown list for validating statistics.
    Prediction Expression Free text to input column expression for validating statistics.
    Model Evaluation and monitoring Select the checkbox to enable feature and prediction drift monitoring.

    This is used for computing statistics after importing model.

    Select to enable model evaluation and performance monitoring.

    If enabled, model will evaluate; otherwise, you can directly approve and then deploy model.

    Metrics Type Select the Metrics Type: default or custom.

    If custom is selected, then user has to upload evaluation files.

    For Python and R Import sheet, set the following properties
    Property Description
    External Id Specifies the model name.
    Tags  
    Model Monitoring Select the checkbox to enable model evaluation and performance monitoring.

    If enabled, model will evaluate; otherwise, you can directly approve and then deploy model.

    Select the checkbox to enable feature and prediction drift monitoring.
    Upload model file Upload control to upload evaluation files.
    Deselecting Enable Model Evaluation does not allow the model version evaluation. For details, see Evaluating a Trained Model.
  7. Select Import Version.
    The imported model displays in the model versions list.