Dataset Templates

Using AnalyticOps, you can create datasets based on dataset templates. This allows you to group all of the datasets related to a use case together in a dataset template. Dataset templates not only help you organize your datasets but also facilitate the dataset creation process by inheriting all the defined parameters. You can create a dataset from a dataset template by either using the default parameters defined for the template or updating them as needed.

The Dataset Template module facilitates the creation and management of dataset templates and datasets. This chapter covers the following details:

Create a Dataset Template

To create a new dataset template:

  1. Open a Project and select Datasets from the Navigation bar.
    The list of Dataset Templates displays in the Work area and the Datasets option highlights in the Navigation bar.


  2. Click the Create Template button.
    The Create Dataset Template dialog displays.

    https://docs.tdaoa.com/images/v6/ug_create_template_dialog.png

  3. In the Step 1 - CATALOG, set the properties:

    Property

    Description

    Name

    Specifies the dataset template name.

    Description

    Specifies the description of the dataset template.

    Feature Catalog Type

    Lets you select a catalog type to define a dataset template. Select ‘Vantage’ for vantage type feature catalog and ‘None’ for no catalog. Depending on this selection, this template dialog morphs the wizard.

    Tags

    Allows you to add tags with the dataset template.

    Database

    Specifies database to store data features information

    Table

    Specifies table to store data features information

  4. In case of Catalog Type as None, following dialog displays:

    https://docs.tdaoa.com/images/v6/ug_create_template_dialog_none.png

    In the Step 2 - CONFIG, set the properties:

    Property

    Description

    Common Settings

    Custom Properties: Lets you define custom properties in the form of key/value pair.
    Template Query: Lets you define a template query that will fetch data for the dataset template.

    Scope Settings

    Allows you to define properties for different scope levels: Train, Evaluate, Score

    Custom Properties: Lets you define custom properties in the form of key/value pair for each of the selected scope.
    Template Query: Lets you define a template query that will fetch data to execute training, evaluation and scoring jobs.

    https://docs.tdaoa.com/images/v6/ug_create_template_dialog_none_config.png

    Click Create to save Dataset Template.

  5. In case of Catalog Type as Vantage, followed by Step 1 - CATALOG, Step 2 - FEATURES displays:

    https://docs.tdaoa.com/images/v6/ug_create_template_dialog_features.png

    Property

    Description

    Query

    Defines SQL query to select features and entity key for catalog.

    Features

    Displays the metadata of the provided query and ability to perform further operations.

    You can define an SQL query to get an introspect on the features fetched directly from database by hitting RUN button. Retrieved results displays in Features table below with the ability to further specify collection by selecting/de-selecting check-boxes adjacent to each feature.

    A small key icon displays in parallel to one and only one row indicating it as Entity Key. You can change entity key by switching between features using the same panel. An entity key cannot be a feature simultaneously, hence upon switching, a prompt appears to confirm selection:

    https://docs.tdaoa.com/images/v6/ug_create_template_features_entity.png

    You can also retrieve actual result of features from SQL query defined earlier by clicking PREVIEW DATA button. Following dialog displays:

    https://docs.tdaoa.com/images/v6/ug_create_template_features_data.png

    By closing this dialog and upon clicking Next in the main dialog, you will be navigated to next step.

  6. In case of Catalog Type as Vantage, Step 3 - ENTITY & TARGET displays:

    https://docs.tdaoa.com/images/v6/ug_create_template_dialog_entityAndTarget.png

    Property

    Description

    Query

    Defines SQL query to select variables (entity sample and targets) for catalog.

    Variables

    Displays the metadata of the provided query and ability to perform further operations.

    You can define an SQL query to retrieve desired variables by hitting RUN button. Retrieved results displays in Variables table below with the ability to further specify sample by selecting/de-selecting check-boxes adjacent to each row. Entity Key selection and operations are exactly the same as FEATURES step.

    You can also retrieve actual result of sample from SQL query and JOIN result of target variables and features defined earlier by clicking PREVIEW DATA button. Following dialog displays:

    https://docs.tdaoa.com/images/v6/ug_create_template_entityAndTarget_result.png

    Navigate to COMBINED QUERY RESULT tab to view combined JOIN SQL query of features and target variables. You can also view its result:

    https://docs.tdaoa.com/images/v6/ug_create_template_entityAndTarget_data.png

    By closing this dialog and upon clicking Next in the main dialog, you will be navigated to next step.

  7. In case of Catalog Type as Vantage, Step 4 - PREDICTIONS displays:

    https://docs.tdaoa.com/images/v6/ug_create_template_dialog_predictions.png

    Property

    Description

    Database Name

    Defines database name to store dataset template

    Table Name

    Defines table name to store dataset template with or without catalog

    Score Query

    Specify SQL Query to load entity keys for scoring

    BYOM Target Column

    Specifies an expression to select BYOM Target Column

    This step also allows you to specify SQL query to load entity keys for scoring.You can also retrieve actual result of entity sample from SQL query and JOIN result of scoring variables and features defined in Step-2 by clicking PREVIEW DATA button. The dialog behavior is exactly same as for entity and target step. You can also define expression for BYOM target column in this step.

  8. Click Create.
    The new dataset template creates with the name specified in the Name field.

Edit a Dataset Template

To edit an existing dataset template:

  1. Select a dataset template from the list.
    The Edit Template button enables.

  2. Click the Edit Template button.
    The Edit Dataset Template dialog depending on the catalog type (Vantage or None) displays accordingly where you can modify the selected template.

    https://docs.tdaoa.com/images/v6/ug_template_edit_dialog.png
  3. After modifying the desired properties, click Update.
    The dataset template saves with all the latest modifications.

Archive a Dataset Template

The archiving feature allows you to hide a dataset template from the list to better organize your templates. You can view the archived templates and unarchive them if required.

Note: The Archiving feature is provided with multiple modules including projects, models, model versions, dataset templates, datasets, and connections.

To archive a dataset template:

  1. Select a dataset template in the list.
    The Actions button enables.

  2. Click the Actions button.
    The Actions menu displays.

    https://docs.tdaoa.com/images/ug_template_archive_action.png
  3. Click Archive Template.
    The dataset template archives and hides from the current list. A confirmation message displays on the top.

  4. To view an Archived template, click the Show Archived option on the top.
    The archived template displays in the list along with an Archived label.

To Un-archive a dataset template:

  1. Select an archived dataset template in the list.
    The Actions button enables.

  2. Click the Actions button.
    The Actions menu displays.

    https://docs.tdaoa.com/images/ug_template_unarchive_action.png
  3. Click Un-Archive Template.
    The dataset template un-archives and the label Archived removes. A confirmation message displays on the top.