Review forecast results

1. Open Runs, or open the experiment that holds the run.
2. Find a run with status Completed and select Results. You can also select Results on the run page.
3. Check the run name, ID and pipeline at the top. Training shows model scores. Prediction shows forecasts. Complete can show both.

Open results from a run. There isn't a separate Results page in the main menu. To find a run, see Find and monitor a run.

1. Choose a Cluster, Series identifier or Target if those choices are shown.
   • A cluster is a group of series.
   • A series identifier picks one series, such as a store and product pair.
   • The target is the value you forecast, such as sales.
   • If there's only one choice, Forecast selects it for you.
2. In Forecast, use Plot to see the pattern or Table to read the values. Use the labels, line styles and markers to tell series apart. Don't rely on color alone.
3. For Complete results, select Metrics to check forecast scores. Training results can also show scores for each cluster or series and the model as a whole. Check which run and series each score refers to.

Each result section loads on its own. For example, you may see model scores while forecast data is still loading. If one section can't load, check its message and use its retry option if shown. Other results may still be ready to use. This also applies to explanations and feature importance results.

• If Results is disabled or shows Results pending, check the run's status in Monitoring. Try again when it's complete. Failed, canceled or incomplete runs don't have completed results to view.
• Results available through API means you can't view these results in Forecast. Ask a technical colleague to get them through the application programming interface (API).
• To download files, open the run page and select Artifacts. You'll need Editor access or above.
  • Search by file name, path or type.
  • Check the full path if files have similar names.
  • Select Download beside the file.
• No artifacts match your search means the search has no matches. Clear it to see all files.
• No artifacts available for this run means no files were returned. A run may not have files yet. Some completed runs also have none.

Follow Compare scenarios to compare changed inputs, or Share runs to give others access.