Dashboard and Widget API
Arize AI for Model Monitoring, Troubleshooting, and Explainability
This feature is currently not generally available and is in development with select design partners.
This API enables you to create dashboards and widgets that can be visualized in 3rd party platforms.
Create a new dashboard in a space for viewing model data.
Note: This query will only return custom created or cloned dashboards. This query will not return preset dashboards.
post
https://api.arize.com
/v1/reporting/create-dashboard
Create Dashboard
curl --request POST 'https://api.arize.com/v1/reporting/created-dashboard' \
--header 'Authorization: API_KEY' \
--data-raw '{
"name":"dashboard_name", //Required
"space_id":"internal_space_name", //Required
}'
This method creates a dashboard in an account. The dashboard can be used to create widgets for data visualization and data export.
This method publishes/adds an already created widget to an already created Dashboard
post
https://api.arize.com
/v1/reporting/publish-widget
Publish Widget
curl --request POST 'https://api.arize.com/v1/reporting/created-dashboard' \
--header 'Authorization: API_KEY' \
--data-raw '{
"widget_ids":["widget-id-1"], //Required
"dashboard_ids":"dashboard-id-1" //Required
}'
This method takes an array of widget ID's (strings) and then connects the widget to the dashboard.
This method creates a timeseries widget that can be performance based or based on data metrics.
post
https://api.arize.com
/v1/reporting/create-timeseries-widget
Create Timeseries Widget
curl --request POST 'https://api.arize.com/v1/reporting/created-timeseries-widget' \
--header 'Authorization: API_KEY' \
--data-raw '{
"metric_type": "EvaluationMetric", //Required
"dashboard_ids":"dashboard-id-1" //Optional
"grid_position": [1,5,4,7], //Optional
"title" : "Hello World" //Optional
}'
This API is designed to allow teams to create a widget object in the platform and attach it to a Dashboard. The widgets can also be created outside of a dashboard and attached to a dashboard at a later point.
post
https://arize.api.com
/v1/reporting/update-timeseries-widget
Plots object:
Parameter | Description | Type | Required |
model_id | The ID of the model to plot | string | Required |
model_version_ids | The version IDs of the model to plot | string | Optional |
dimension_metric | The name of the dimension to plot for data metrics, not used for evaluation metrics. Feature, prediction or actual | string | Optional |
evaluation_metric | The evaluation metric. Supported values: "accuracy", "precision", "recall", "f1", "sensitivity", "specificity", "false_positive_rate", "false_negativerate", "AUC", "log_loss", "mae", "mape", "rmse", "mse" | string | Optional |
filters | Filters object to apply to widget* (ask Arize team for latest filter query examples) | dict/obj | Optional |
aggregation | The aggregation type for data metrics: "average", "percent", "count", "unique", "std', "min", "max" | string | Required |
curl --request POST 'https://api.arize.com/v1/reporting/update-timeseries-widget' \
--header 'Authorization: API_KEY' \
--data-raw '{
"metric_type": "EvaluationMetric", //Required
"dashboard_ids":"dashboard-id-1" //Optional
"grid_position": [1,5,4,7], //Optional
"title" : "Hello World" //Optional
"plots" : {
"model_id" : "model-1" //Required
"model_version_ids" : [] //Optional
"dimension_metric" : null //Optional
"evaluation_metric" : "accuracy" //Optional
"filters" : null //Optional
"model_version_ids" : ["v1"] //Optional
"aggregation" : null . //Optional
}
}'
The update time series API is designed to allow teams to set the data visualization calculations and options of widgets on dashboards.
Last modified 5mo ago