CODEWITHSUNDEEP
pythonpandasdataframeindexingnan
PDiracDelta
PDiracDelta

Reputation: 2558

cannot set pandas column values using series, sets everything to np.nan instead

I have the following pandas (pd) dataframe:

> df = pd.DataFrame({'x':[1,2,3], 'y':[4,5,6], 'z':[7,8,9]}, index=['one', 'two', 'three'])
> df
       x  y  z
one    1  4  7
two    2  5  8
three  3  6  9

and a Series:

s = pd.Series([99,99,99])

When I try to assign these values in b to some column in df, I don't get any error but all values in that column are set to nan instead: 

> df['y'] = s
> df
       x   y  z
one    1 NaN  7
two    2 NaN  8
three  3 NaN  9

I have set dataframe column values many times before using this assignment technique, why doesn't this work (anymore)?

Upvotes: 0

Views: 2627

Answers (2)

wwnde
wwnde

Reputation: 26676

ICCU:

Another way of doing it. Please try df.assign

df.assign(y=s.values)

enter image description here

Upvotes: 0

PDiracDelta
PDiracDelta

Reputation: 2558

The issue is with the index values in the DataFrame and Series: they don't match. By default, any pandas index has integer values which count up from 0, so if you don't modify them and the lengths of your column and Series match, there is no problem.

However, you modified the index values of df and set them to ['one', 'two', 'three']. You should make sure that:

  • either the Series uses the same index as the DataFrame:

    > s = pd.Series({'one': 99, 'two': 99, 'three': 99})
> df['y'] = s

  • or, you can just use the (index-less) values in s:

    > df['y'] = s.values

to obtain:

> df
       x   y  z
one    1  99  7
two    2  99  8
three  3  99  9

Upvotes: 2

Related Questions