Building Hierarchies with Group By Columns: A Comparison of PySpark and Pandas Approaches