ENH: extend DataFrame.insert to also insert rows

Is your feature request related to a problem?

I wish I could use pandas to directly insert a row at a specific position in a data frame, just like I can insert a column via the DataFrame.insert method. This would be similar to the add_row function that exist for R data frames.

Describe the solution you’d like

expand the functionality of DataFrame.insert to take an axis argument and allow for the insertion of rows when axis is set to 0.

API breaking implications

as insert currently inserts a column by default, the default argument to axis should be set to 1. When this is done, I don’t foresee any breaks.

Describe alternatives you’ve considered

An alternative is to use concatenate. This is tedious though, because it requires several concat operations (upper block + row to insert + lower block) and error prone. Another alternative (not in-place) is to use np.insert like

import seaborn as sns
import pandas as pd
import numpy as np

def insert_row(df, row, position):
    out = pd.DataFrame(np.insert(arr=df.values,
        	                     obj=position,
                	             values=row,
                                 axis=0),
           	            columns=df.columns)

    return out
iris = sns.load_dataset("iris")
row_to_insert = iris.iloc[1, :]
insert_row(iris, row_to_insert, 1)

Additional context

Code example to insert column:

import seaborn as sns
iris = sns.load_dataset("iris")
col_to_insert = iris.iloc[:, 1]
position_to_insert = 1
iris.insert(position_to_insert, "new_column", col)

Desired behaviour:

row_to_insert = iris.iloc[1, :]
iris.insert(position_to_insert, "new_row", row, axis=0)

[rhshadrach]

Agreeing with @TomAugspurger, while the concat option might be a bit, it truly reflects what’s going on. If users really need this, it is easy to write a function that (I think) would be just as performant as what we could do internally.

[MarcoGorelli]

closing as there doesn’t seem to be much support for this