CODEWITHSUNDEEP
pandasdataframe
Neil Traft
Neil Traft

Reputation: 19726

Iterate over "entries" of a pandas DataFrame

Say I have a (potentially multi-indexed, but maybe not) dataframe. For example:

iterables = [['foo', 'bar'], ['one', 'two']]
idx = pd.MultiIndex.from_product(iterables, names=['first', 'second'])
df = pd.DataFrame(np.random.randn(4, 2), index=idx, columns=('metric1', 'metric2'))

result:

               metric1   metric2
first second                    
foo   one     0.189589  0.787533
      two     0.176290 -0.080153
bar   one     0.077977 -0.384613
      two     0.658583  0.436177

There are many ways to iterate over each element in this dataframe, but most of them involve two nested for-loops like:

for r in df.index:
    for c in df.columns:
        print r, c, df.loc[r,c]

which yields

('foo', 'one') metric1 -0.00381142017312
('foo', 'one') metric2 -0.755465118408
('foo', 'two') metric1 0.444271742766
('foo', 'two') metric2 0.18390288873
('bar', 'one') metric1 0.512679930964
('bar', 'one') metric2 -0.134535924251
('bar', 'two') metric1 1.93222192752
('bar', 'two') metric2 0.609813960012

Is there a way to do that in one loop (such that I have access to row name(s) and column name(s) for each element as I'm iterating)? If it's only possible with a regular Index I'd still be interested.

Upvotes: 0

Views: 34

Answers (1)

akuiper
akuiper

Reputation: 215117

You can stack the data frame as a Series and then loop it in one go:

for ind, val in df.stack().items():
    print(ind, val)

('foo', 'one', 'metric1') -0.752747101421
('foo', 'one', 'metric2') 0.318196702146
('foo', 'two', 'metric1') -0.737599211438
('foo', 'two', 'metric2') -1.08364260415
('bar', 'one', 'metric1') 1.87757917778
('bar', 'one', 'metric2') -2.29588862481
('bar', 'two', 'metric1') -0.301414352794
('bar', 'two', 'metric2') 0.610076176389

Upvotes: 1

Related Questions