python - iterate over pandas.DataFrame with MultiIndex by index names -

i iterate on dataframe multiindex while using names access specific value of index column. example, given following

import pandas pd index = pd.multiindex.from_product([range(2), range(3)], names=['index_a', 'index_b']) table = pd.dataframe({'my_column': range(len(index))}, index=index) 

i iterate on rows of table using code like:

for row in named_index_iterator(table):     print(row.my_column, row.index_a, row.index_b) 

or

for row in named_index_iterator(table):     print(row.my_column, row.index.index_a, row.index.index_b) 

to implement named_index_iterator cannot use itertuples or iterrows in dataframe gives plain tuples, not named tuples, index. cannot use like:

for data_row, index_row: itertools.zip_longest(table.itertuples(), table.index):  

as iterator on table.index gives again plain tuples, not named tuples.

as workaround use

for row in table.reset_index().itertuples(): 

but copies table.

answering own question references.

i created following utility iterate on indexes names:

import collections  def df_iter_with_index_names(table):     indexnames = collections.namedtuple('indexnames', table.index.names)     row in table.itertuples():         yield (indexnames(*row.index), row) 

with usage like:

import collections import pandas pd index = pd.multiindex.from_product([range(2), range(3)], names=['index_a', 'index_b']) table = pd.dataframe({'my_column': range(len(index))}, index=index) print(table)  def df_iter_with_index_names(table):     indexnames = collections.namedtuple('indexnames', table.index.names)     row in table.itertuples():         yield (indexnames(*row.index), row)  index, row in df_iter_with_index_names(table):     print(index.index_a, row.my_column) 

it can improved remove index name row tuple, leftover dataframe.itertuples(), can live that.