I have a pyspark.pandas.frame.DataFrame object (that I called from `pandas_api` on a pyspark.sql.dataframe.DataFrame object).  I have a complicated transformation that I would like to apply to this data, and in particular I would like to apply it in blocks based on the value of a column 'C'.

If it had been a pandas.core.frame.DataFrame object, I could do:

    for _,chunk in df.groupby("C"):

        // do stuff

When I try this with a pyspark.pandas.frame.DataFrame object, I get `KeyError: (0,)`.  

My question is: how do I get access to the grouped data in a pyspark.pandas.groupby.DataFrameGroupBy object? Is this possible at all, or am I only allowed to run aggregate functions?