CODEWITHSUNDEEP
pythonpandasdataframe
Arun Nalpet
Arun Nalpet

Reputation: 1300

python pandas "cannot set a row with mismatched columns" error

I am facing an issue while adding a new row to the data set.

Here is the example DataFrame.

column_names = ['A','B','C']
items = [['a1','b1','c1'],['a2','b2']]
newDF = pd.DataFrame(items,columns=column_names)
print(newDF)

output:

    A   B     C
0  a1  b1    c1
1  a2  b2  None

Since c2 was missing, it was replaced with None. This is fine and as expected.

Now if i continue to add similar rows to this existing DataFrame, like this: 

newDF.loc[len(newDF)] = ['a3','b3']

I get the error "cannot set a row with mismatched columns".

How can I add this additional row, so that it will automatically take care of missing c3 with None or NaN?

Upvotes: 15

Views: 42091

Answers (3)

Karn Kumar
Karn Kumar

Reputation: 8816

what about just :

>>> print(newDF)
    A   B     C
0  a1  b1    c1
1  a2  b2  None
>>> newDF
    A   B     C
0  a1  b1    c1
1  a2  b2  None

Just place new index 2 with new values a3 & b3 and last column.

>>> newDF.loc['2'] = ['a3','b3', np.nan]
>>> newDF
    A   B     C
0  a1  b1    c1
1  a2  b2  None
2  a3  b3   NaN

OR

>>> row = ['a3','b3', np.nan]
>>> newDF.loc['2'] = row
>>> newDF
    A   B     C
0  a1  b1    c1
1  a2  b2  None
2  a3  b3   NaN

Another way around: appending to Dataframe, the new values across the row for desired columns as we have for A & B this another column for them row will become NaN

>>> row
['a3', 'b3']
>>> newDF.append(pd.DataFrame([row],index=['2'],columns=['A', 'B']))
    A   B     C
0  a1  b1    c1
1  a2  b2  None
2  a3  b3   NaN

Upvotes: 4

Sruthi
Sruthi

Reputation: 3018

You specify your new row as a dictionary and create a dataframe out of it.

new_entry = {'A': ['a3'], 'B': ['b3']}
new_entry_df=pd.DataFrame.from_dict(new_entry)

Now this can be appended to the original dataframe

newDF.append(new_entry_df)

    A   B     C
0  a1  b1    c1
1  a2  b2  None
0  a3  b3   NaN

Upvotes: 3

timgeb
timgeb

Reputation: 78690

One option is DataFrame.append:

>>> new_row = ['a3', 'b3']
>>> newDF.append(pd.Series(new_row, index=newDF.columns[:len(new_row)]), ignore_index=True)
    A   B     C
0  a1  b1    c1
1  a2  b2  None
2  a3  b3   NaN

Upvotes: 10

Related Questions