[UX] Test and Start Forecast

[xeniatup]

[UX META] OpenSearch Dashboards Forecasting Dev Ops Metrics
[Proposal] Forecasting Dev Ops Metrics and Financial/Inventory Metrics

What solution would you like?

Forecast page structure

After creating a forecast user navigates to the forecast page. Forecast page serves “View” and “Edit” experiences for the forecast.
The header of the forecast page shows the name of the forecast, the state, and the available actions (depending on the state). The list of the actions in correspondence with the state is covered in “List Forecasts”.

OUI: OuiHealth, OuiDescriptionList, OuiTabs, OuiDescribedFormGroup, OuiBottomBar

User can test forecast configuration (data source and model parameters) and evaluate a sample forecasting result by going into Test flow (“Initializing test” state → “Test complete” state). It is a primary way to proceed after creating a forecast. The forecast goes into “Test complete” state automatically after completing a test.

Screen 3.1: Forecast in “Initializing test” state after being created

Screen 3.2: Forecast in “Test complete” state.

Test cycle runs the forecast once (Initializing test state) and stops (Test complete state) to predict a set of datapoints for each interval going one horizon into the future and a set of historical forecasts to allow user to evaluate the model output before setting it up to run continuously.
After the test forecast has been completed user can visually inspect the forecasting results on a graph and either change the configuration and repeat the test or start forecasting continuously.

The visualization graph controls are described in “View Forecasting Results” and editing forecast configuration is covered in “Edit forecast”

User selects “Start forecast” to move it to “Initialize forecast” state. After initializing the forecast automatically transitions into “Running” state. In the “Running” state the visualization graph is updating automatically on a frequency defined by forecasting interval.

Screen 3.3 Forecast running