Note that the dataframes in the partitioned_data list is dynamically populated with data at runtime, although the size of the list is statically determined at compile time. I doubt map_task will work in this case.

it should