calculating averages of multiple columns, ignoring NaN pandas numpy

Question

I have a basic table of values:

import pandas as pd
import numpy as np
test = pd.read_csv('mean_test.csv')
test.replace('n/a',np.nan)
test


value1  value2  value3
1   9   5
5   NaN 4
9   55  NaN
NaN 4   9

I want to work out the average of the three values, ignoring NaN, so for the second row it would be (5+4)/2. Therefore I can't use the .replace function to put a zero in NaN's place. I have searched through some other questions, but can't find anything that covers this. Am I missing something obvious?

Answer 1

Pandas takes care of the NaN for you:

>>> df
value1  value2  value3
0       1       9       5
1       5     NaN       4
2       9      55     NaN
3     NaN       4       9

>>> df.mean(axis=1)
0     5.0
1     4.5
2    32.0
3     6.5
dtype: float64