通过计算熊猫另一列中的不同值来创建新列

Question

Hello I have a dataframe such as:

COL1_1 COL1_3            COL2
Chr1_0 Canis_lupus       A
Chr1_0 Canis_lupus       A
Chr1_0 Canis_lupus       B
Chr1_0 Canis_lupus       B
Chr1_0 Canis_lupus       B
Chr1_0 Felis_cattus      B
Chr1_0 Felis_cattus      B
Chr2_0 Felis_cattus      A
Chr2_0 Felis_cattus      B
Chr2_1 Felis_cattus      C
Chr2_1 Felis_cattus      D
Chr2_1 Felis_cattus      E

and the idea is within each COL1_1 and COL1_3 count the number of distinct COL2 .

ex : for Chr1_0 and Canis_lupus there are 2 distinct COL2 (A and B), so I put 2 into the new COL3 .

if there is only one value, I put a 0.

here I should then get

COL1_1 COL1_3            COL2  COL3
Chr1_0 Canis_lupus       A     2
Chr1_0 Canis_lupus       A     2
Chr1_0 Canis_lupus       B     2
Chr1_0 Canis_lupus       B     2
Chr1_0 Canis_lupus       B     2
Chr1_0 Felis_cattus      B     0
Chr1_0 Felis_cattus      B     0
Chr2_0 Felis_cattus      A     2
Chr2_0 Felis_cattus      B     2
Chr2_1 Felis_cattus      C     3
Chr2_1 Felis_cattus      D     3
Chr2_1 Felis_cattus      E     3

maybe an idea would be to groupby (COL1_1 and COL1_3`) and count number of distinct COL2 values.

Answer 1

Use GroupBy.transform with DataFrameGroupBy.nunique and Series.mask for replace 1 to 0 :

df['COL3'] = (df.groupby(['COL1_1', 'COL1_3']).COL2.transform('nunique')
                .mask(lambda x: x == 1, 0))

Or use replace :

df['COL3'] = df.groupby(['COL1_1', 'COL1_3']).COL2.transform('nunique').replace({1:0})

---

print (df)
    COL1_1        COL1_3 COL2  COL3
0   Chr1_0   Canis_lupus    A     2
1   Chr1_0   Canis_lupus    A     2
2   Chr1_0   Canis_lupus    B     2
3   Chr1_0   Canis_lupus    B     2
4   Chr1_0   Canis_lupus    B     2
5   Chr1_0  Felis_cattus    B     0
6   Chr1_0  Felis_cattus    B     0
7   Chr2_0  Felis_cattus    A     2
8   Chr2_0  Felis_cattus    B     2
9   Chr2_1  Felis_cattus    C     3
10  Chr2_1  Felis_cattus    D     3
11  Chr2_1  Felis_cattus    E     3