CODEWITHSUNDEEP
pythonpandas
AZhao
AZhao

Reputation: 14415

Distinct combinations values in Pandas DataFrames

Is there an easy way to pull out the distinct combinations of values in a dataframe? I've used pd.Series.unique() for single columns, but what about multiple columns?

Example data:

df = pd.DataFrame(data=[[1, 'a'], [2, 'a'], [3, 'b'], [3, 'b'], [1, 'b'], [1, 'b']], 
                  columns=['number', 'letter'])

Expected output:
(1, a)
(2, a)
(3, b)
(1, b)

Ideally, I'd like a separate Series object of tuples with the distinct values.

Upvotes: 5

Views: 1919

Answers (2)

EdChum
EdChum

Reputation: 394041

You can set the index to those columns and then call unique on the index:

In [165]:
idx = df.set_index(['number','letter']).index
idx.unique()

Out[165]:
array([(1, 'a'), (2, 'a'), (3, 'b'), (1, 'b')], dtype=object)

Upvotes: 5

Alexander
Alexander

Reputation: 109546

You can zip the columns and create a set:

>>> set(zip(df.number, df.letter))
{(1, 'a'), (1, 'b'), (2, 'a'), (3, 'b')}

Upvotes: 7

Related Questions