GroupByDocs
Groups the rows of this DataFrame based on the values in one or more specified cols. Each unique value in a key column — or a unique tuple of values for multiple columns — defines the group consisting of all rows where the column(s) contain that value combination.
Returns a GroupBy — a dataframe-like structure that contains all unique combinations of key values along with the corresponding groups of rows (each represented as a DataFrame) as rows.
A GroupBy can then be:
transformedinto a newGroupBy;reducedinto aDataFrame, where each group is collapsed into a single representative row;aggregatedinto aDataFrame, where each group is transformed into one or more rows of derived values;pivotedinto aPivotGroupBystructure, which combinespivotandgroupByoperations and then reduced or aggregated into aDataFrame.
Grouping keys can also be created inline (i.g. by creating a new column using expr or simply renaming the old one using named):
// Create a new column "newName" based on existing "oldName" values
// and use it as a grouping key:
df.groupBy { expr("newName") { oldName.drop(5) } }Check out Grammar.
This can include column groups and nested columns.
See Selecting Columns.
For more information: See groupBy on the documentation website.
Don't confuse this with group, which groups columns into a column group.
Types
Provides predefined shortcuts for the most common statistical aggregation operations that can be applied to each group within a GroupBy.
Selecting columns for various DataFrame operations can be done in the following ways: