Pandas Indexing Using Multiple Isin Clauses

If I want to do is-in testing on multiple columns at once, I can do: >>> from pandas import DataFrame >>> df = DataFrame({'A': [1, 2, 3], 'B': [1, 4, 7], 'C' : [

Solution 1:

You could put both the isin conditions in &

df[df['A'].isin([1, 3]) & df['B'].isin([4, 7, 12])]
   AB   C
23718

You could also use query function like

c_a = [1, 3]
c_b = [4, 7, 12]
df.query('(B in @c_b) & (A in @c_a)')

   A  B   C
2  3  7  18

Solution 2:

TBH, your current approach looks fine to me; I can't see a way with isin or filter to improve it, because I can't see how to get isin to use only the columns in the dictionary or filter to behave as an all.

I don't like hardcoding column names, though, so I'd probably write this as

>>>keep = {'A': [1, 3], 'B': [4, 7, 12]}>>>df[df[list(keep)].isin(keep).all(axis=1)]
   A  B   C
2  3  7  18

or with .loc if I needed a handle.

Solution 3:

You could put both conditions in as a mask and use &:

In[12]:

df[(df['A'].isin([1,3])) & (df['B'].isin([4,7,12]))]
Out[12]:
   ABC23718

Here the conditions require parentheses () around them due to operator precedence

Slightly more readable is to use query:

In[15]:

df.query('A in [1,3] and B in [4,7,12]')
Out[15]:
   ABC23718