CODEWITHSUNDEEP
pythonpandasindexingseries
Sebastian Allard
Sebastian Allard

Reputation: 48

Adding values of two Pandas series with different column names

I have two pandas series of the same length but with different column names. How can one add the values in them?

series.add(other, fill_value=0, axis=0) does avoid NaN-values, but the values are not added. Instead, the result is a concatenation of the two series.

Is there a way to obtain a new series consisting of the sum of the values in two series?

Upvotes: 2

Views: 1622

Answers (2)

jpp
jpp

Reputation: 164623

Mismatched indices

This issue is your 2 series have different indices. Here's an example:

s1 = pd.Series([1, np.nan, 3, np.nan, 5], index=np.arange(5))
s2 = pd.Series([np.nan, 7, 8, np.nan, np.nan], index=np.arange(5)+10)

print(s1.add(s2, fill_value=0, axis=0))

0     1.0
1     NaN
2     3.0
3     NaN
4     5.0
10    NaN
11    7.0
12    8.0
13    NaN
14    NaN
dtype: float64

You have 2 options: reindex via, for example, a dictionary or disregard indices and add your series positionally.

Map index of one series to align with the other

You can use a dictionary to realign. The mapping below is arbitrary. NaN values occur where, after reindexing, values in both series are NaN:

index_map = dict(zip(np.arange(5) + 10, [3, 2, 4, 0, 1]))
s2.index = s2.index.map(index_map)

print(s1.add(s2, fill_value=0, axis=0))

0     1.0
1     NaN
2    10.0
3     NaN
4    13.0
dtype: float64

Disregard indices; use positional location only

In this case, you can either construct a new series with the regular pd.RangeIndex as index (i.e. 0, 1, 2, ...), or use an index from one of the input series:

# normalized index
res = pd.Series(s1.values + s2.values)

# take index from s1
res = pd.Series(s1.values + s2.values, index=s1.index)

Upvotes: 3

shx2
shx2

Reputation: 64298

The values attribute lets you access the underlying raw numpy arrays. You can add those.

raw_sum = series.values + other.values
series2 = Series(raw_sum, index=series.index)

This also works:

series2 = series + other.values

Upvotes: 1

Related Questions