How to calculate column mean and row skip non numeric and na

[Mekala]

Hi,
I want to calculate column mean and row mean & skip "na" and "non-numeric". I use below code, but it gives some warning as show below:

__main__:5: FutureWarning: convert_objects is deprecated. Use the data-type specific converters pd.to_datetime, pd.to_timedelta and pd.to_numeric.

import pandas as pd df = pd.read_csv(r"D:\Data\PythonCodes\outsummary2.csv") print((df[['s2']].convert_objects(convert_numeric=True)).mean(skipna=True))

my csv data:

s1	s2	s3 A	1	2	1 B	na	5	3 U	1	na	0 Z	0 Z	0	2	2

is there any efficient way to skip "non-numeric" and "na", while calculating column mean (or sum)

[anbu23]

pd.to_numeric(df['s2'], errors='coerce').mean()

[Mekala]

It works, Thanks a lot. Can we get the row index of actual numeric values?

[anbu23]

>>> df s1 0 1 1 na 2 1 3 0 4 0 >>> for i, num in df['s1'].iteritems(): ... if type(num) is int: ... print(i) ... 0 2 3 4

[Mekala]

import pandas as pd df = pd.read_csv(r"D:\Data\outsummary2.csv") df2 = df.convert_objects(convert_numeric=True) for i, num in df2['s2'].iteritems(): print(num, type(num)) if type(num) is int: print(i)

Output:
it always float type.
2.0 <class 'numpy.float64'>
5.0 <class 'numpy.float64'>
nan <class 'numpy.float64'>
nan <class 'numpy.float64'>
2.0 <class 'numpy.float64'>

[anbu23]

>>> import numpy as np >>> df2 0 2.0 1 5.0 2 NaN 3 2.0 Name: s2, dtype: float64 >>> for i, num in df2.iteritems(): ... if not np.isnan(num): ... print(num) ... 2.0 5.0 2.0