MergeError using pandas

Question

df1 = pd.DataFrame(columns=["x","y","z"])
df2 = pd.DataFrame({"x":[1],"y":[2]}, index=["foo"])

I'm trying to get this result:

       z  x  y
foo  NaN  1  2

I tried

df1.merge(df2,how="outer", on=["x","y"], right_index=True)

which gives the error

MergeError: Can only pass argument "on" OR "left_index" and "right_index", not a combination of both.

However, neither

df1.merge(df2,how="outer", on=["x","y"])

nor

df1.merge(df2,how="outer", left_index=True, right_index=True)

give the desired result...

Your first empty DataFrame is fairly pointless. I think you're looking to reindex: df2.reindex(['z', 'x', 'y'], axis=1) — ALollz
– ALollz, Commented Mar 31, 2021 at 20:05
Thanks @ALollz. I need df1 because it is not always empty. It depends on the user's input. — HappyPy
– HappyPy, Commented Mar 31, 2021 at 20:47

Krzysztofik · Accepted Answer · 2021-03-31 20:59:49Z

1

The following command:

    df2.merge(df1, on=['x', 'y'], how='outer')

produces

   x  y    z
0  1  2  NaN

The index is not the one you need, so you can do this instead:

    df2.reset_index().merge(df1, on=['x', 'y'], how='outer').set_index('index')

producing

       x  y    z
index           
foo    1  2  NaN

If you want to remove the word 'index', name the above dataframe df3 and:

    del df3.index.name

resulting in

     x  y    z
foo  1  2  NaN

answered Mar 31, 2021 at 20:31

Krzysztofik

556 bronze badges

Sign up to request clarification or add additional context in comments.

Comments

jcaliz · Accepted Answer · 2021-03-31 21:20:56Z

0

Why don't just:

df2['z'] = df1['z']

answered Mar 31, 2021 at 21:20

jcaliz

4,0732 gold badges11 silver badges16 bronze badges