CODEWITHSUNDEEP
pythonpandasnumpydataframe
Wouter Vandenputte
Wouter Vandenputte

Reputation: 2103

Python dataframe vector norm efficient

I'd like to calculate for each row of a dataframe the norm (i.e. the length of the vector in euclidean distance). I know functions like mean() and sum() exist but no norm(). So I tried implementing it myself by doing

df.apply(lambda values: math.sqrt(sum([v**2 for v in values])), axis=1)

but this is realy slow compared to e.g. the sum function. Is there a simple (and fast) pandas implementation for it?

Upvotes: 1

Views: 782

Answers (1)

juanpa.arrivillaga
juanpa.arrivillaga

Reputation: 96236

Sure, use numpy here:

In [1]: import pandas as pd, numpy as np

In [2]: df = pd.DataFrame(data=[[1,2,3],[4,5,6]], columns=['a','b','c'])

In [3]: np.linalg.norm(df, axis=1)
Out[3]: array([3.74165739, 8.77496439])

Upvotes: 2

Related Questions