I have the following data frame:
import pandas as pd
df = pd.DataFrame({'AAA' : ['w','x','y','z'], 'BBB' : [10,20,30,40],'CCC' : [100,50,-30,-50]})
Which looks like this:
In [32]: df
Out[32]:
AAA BBB CCC
0 w 10 100
1 x 20 50
2 y 30 -30
3 z 40 -50
What I want to do is to perform function operation on every row for every column except those with non-numerical value (in this case AAA). In the real case the non-numerical case is always on first column, and the rest (could be greater than 2 columns) are always numerical.
The final desired output is:
AAA BBB CCC Score
0 w 10 100 110
1 x 20 50 70
2 y 30 -30 0
3 z 40 -50 -10
I tried this but failed:
import numpy as np
df["Score"] = df.apply(np.sum, axis=1)
What's the right way to do it?
Update2:
This is the code that give SettingWithCopyWarning. Please fresh start the ipython for testing.
import pandas as pd
import numpy as np
def cvscore(fclist):
sd = np.std(fclist)
mean = np.mean(fclist)
cv = sd/mean
return cv
def calc_cvscore_on_df(df):
df["CV"] = df.iloc[:,1:].apply(cvscore, axis=1)
return df
df3 = pd.DataFrame(np.random.randn(1000, 3), columns=['a', 'b', 'c'])
calc_cvscore_on_df(df3[["a","b"]])
To select everything but the first column, you could use df.iloc[:, 1:]:
In [371]: df['Score'] = df.iloc[:, 1:].sum(axis=1)
In [372]: df
Out[372]:
AAA BBB CCC Score
0 w 10 100 110
1 x 20 50 70
2 y 30 -30 0
3 z 40 -50 -10
To apply an arbitrary function, func, to each row:
df.iloc[:, 1:].apply(func, axis=1)
For example,
import numpy as np
import pandas as pd
def cvscore(fclist):
sd = np.std(fclist)
mean = np.mean(fclist)
cv = sd/mean
return cv
df = pd.DataFrame({'AAA' : ['w','x','y','z'], 'BBB' : [10,20,30,40],
'CCC' : [100,50,-30,-50]})
df['Score'] = df.iloc[:, 1:].apply(cvscore, axis=1)
print(df)
yields
AAA BBB CCC Score
0 w 10 100 1.211386
1 x 20 50 0.868377
2 y 30 -30 NaN
3 z 40 -50 -5.809058
Collected from the Internet
Please contact [email protected] to delete if infringement.
Comments