Adding rows to a multi-indexed DataFrame

Asked Jan 05 '20 at 18:10

Active Feb 20 '20 at 07:25

Viewed 1,529 times

I want to add/update rows to a Pandas DataFrame which is multi-indexed, and the question Add one row to pandas DataFrame does not seem to address this problem. For instance, the following:

import pandas as pd


df = pd.DataFrame(columns=['x', 'y', 'z', 't'])
df.set_index(['x', 'y'])
df.loc[(1, 2), :] = [1, 3]

gives the error

KeyError: "None of [Int64Index([1, 2], dtype='int64')] are in the [index]"

By the way: is there a simpler way to initialize the DataFrame than the one above ? I tried

df = pd.DataFrame(columns=['z', 't'], index=pd.MultiIndex(names=['x', 'y']))

but this gives the error

TypeError: Must pass both levels and codes

Thanks.

edited Feb 20 '20 at 07:25

Martin Gergov

1,556
4
20
29

asked Jan 05 '20 at 18:10

user0

Use `df.set_index(['x', 'y'], inpace=True)` – jezrael Jan 05 '20 at 18:15
1

Or `df = df.set_index(['x', 'y'])` – jezrael Jan 05 '20 at 18:15
1

Thanks for this answer! – user0 Jan 05 '20 at 19:01
Just fyi, better use @jezrael second comment, `inplace` argument is not recommended by pandas devs: see [this answer](https://stackoverflow.com/a/59242208/4909087) of mine which links to a relevant GitHub issue – cs95 Jan 05 '20 at 19:59

Adding rows to a multi-indexed DataFrame

0 Answers0