CODEWITHSUNDEEP
pythonpandasdataframe
user1823812
user1823812

Reputation: 199

Grouping dataframes in pandas?

I'm pretty new to python pandas and working with dataframes. Lets say that I have the dataframe shown below :

A B C

3 2 3
4 2 4
3 2 1 
5 6 6

I want to find how many rows in my dataframe have the same value in the A and B column and for these rows I want to store the C value. So for example in this dataframe, I want to store the 1st and 3rd C value, since the A and B in each of these rows have the same values. Basically I want a way to print something like: "For A = 3 and B = 2 the possible C values are : 3,1" and find those pairs. Im following the official pandas documentation but I can't seem to find this.

Upvotes: 4

Views: 107

Answers (3)

Andy Hayden
Andy Hayden

Reputation: 375377

You can use the unique SeriesGroupBy method:

In [11]: df.groupby(["A", "B"])["C"].unique()
Out[11]:
A  B
3  2    [3, 1]
4  2       [4]
5  6       [6]
Name: C, dtype: object

See also nunique to get the number of unique elements.

Upvotes: 2

Joseph
Joseph

Reputation: 1136

You may try this:

In [187]: df
Out[187]: 
   A  B  C
0  3  2  3
1  4  2  4
2  3  2  1
3  5  6  6

In [188]: df[df.groupby(['A', 'B']).transform(np.count_nonzero).C>=2]
Out[188]: 
   A  B  C
0  3  2  3
2  3  2  1
Name: C, dtype: float64

Upvotes: 2

reptilicus
reptilicus

Reputation: 10397

Not sure if I follow but this might get you going:

df = DataFrame({"a": [3,4,3,5], "b":[2,2,2,6], "c": [3,4,1,6]})
In [38]: for i, g in df.groupby(("a", "b")):
             print  i, g["c"].values
         ....:
(3, 2) [3 1]
(4, 2) [4]
(5, 6) [6]

Upvotes: 2

Related Questions