Now get additional details/metrics around all your algorithm runs with AutoML
Until today, when using AutoML in Forecast customers could only determine the winning algorithm. Although useful, this did not give customers transparency into all the model runs.
Until today, when using AutoML in Forecast customers could only determine the winning algorithm. Although useful, this did not give customers transparency into all the model runs.
With today’s change, not only will customers be able to see the winning algorithm but also get metrics for all algorithms that succeded as well as any error messages for those did not. The former is useful for customers who might want to optimize for a metric that is different from the one AutoML has chosen and the latter to make updates to the imported data to ensure a successful run across all algorithms if so desired. These can be viewed from the GetAccuracyMetrics and DescribePredictor APIs. As with all features, this can be accessed via the console or AWS CLI.