Customizing Dashboard in the platform

Add the report to dashboard

Once custom-reports are created, we need to configure how/where the reports need to be shown in the User Interface. i.e. in which tab of the dashboard, what height should be used, etc. This is done in the api_config.py with the CUSTOM_REPORT_TABS setting.

CUSTOM_REPORT_TABS Structure

CUSTOM_REPORT_TABS__Model = [
    {
        'text': 'Model Dependent',
        'data': [
            {'name': 'histogram_distribution', 'width': '33%', 'layout': {'height': 300}},
            {'name': 'summary_stats', 'width': '33%', 'layout': {'height': 300}},
        ],
    },
]

The CUSTOM_REPORT_TABS setting can be split up into three parts:

Object Level

CUSTOM_REPORT_TABS__<<OBJECT_NAME>> = ...

The name of the setting contains which dashboard the structure being described is for. For example, CUSTOM_REPORT_TABS__Model is for the dashboard of Models in Model Registry. Currently in the platform, the following Dashboards exist:

• DataElement
• DataQuality
• Feature
• Model
• Dataset
• Experiment
• Policy

Dashboard >> Tab Level

{
    'text': 'Model Dependent',  # The name of the tab
    'data': [],  # The reports to display within the tab
}

The next level is within these dashboards, which tabs need to be shown. Tabs are a way to categorize and segregate the reports into a logical structure.

• text is used to specify the tab name for display
• data is a list of reports to display under the tab

Dashboard >> Tab >> Report Level

    {'name': 'histogram_distribution', 'width': '33%', 'layout': {'height': 300}}

The next level represents a single report.

• name is the name property in the report class - which is used to match it to this item

• width / height (optional) can be used to customize the dimensions of the report

• The accepted values for width are:

  • 25%
  • 33%
  • 50%
  • 100%

  If width parameter is not provided, it defaults to 100%

To summarize, the CUSTOM_REPORT_TABS structure above shows that in the platform, the Model dashboard includes one Model Dependent tab, which contains 2 custom reports: histogram_distribution and summary_stats

Notes:

1. The platform comes with a set of default reports, which are set using CUSTOM_REPORT_TABS. These can be customized as required.
2. This configuration is used for all job-types: simulation, comparison, and validation.