the mean is only calculated for a single column.
According to the docs (https://h2o-release.s3.amazonaws.com/h2o/rel-vajda/1/docs-website/h2o-docs/data-munging/groupby.html) I would expect:
In the first case all columns except the column grouped upon get their mean calculated.
In the second case all columns specified get their mean calculated.
In the example below, I would expect the mean for all columns grouped by airport: