replace all floats in df with corresponding index name-CodePudding

I want to replace all values in my df that are float (excluding nans), with the name of the index of the corresponding row.

I have this:

index1                10.0                          190.6   
index2                17.9                          NaN   
index3                NaN                           8.0
index4                9.0                           70.0

I want to have this:

index1                index1                        index1                
index2                index2                        NaN   
index3                NaN                           index3                
index4                index4                        index4

Any ideas?

CodePudding user response：

use:

df2=df.T
'''
        index1  index2  index3  index4
col1    10.0    17.9    nan     9.0
col2    190.6           8.0     70.0

'''
df2=df2.replace(np.nan,'nan') #Since nan values are float type, I convert them to string nan.

for i in df2.columns:
        df2[i]=df2[i].apply(lambda x: i if isinstance(x, float) else x)

df2=df2.T.replace('nan',np.nan) #convert string nans to real nans
print(df2)
'''
index1  index1  index1
index2  index2     NaN
index3     NaN  index3
index4  index4  index4
'''

CodePudding user response：

Technically, np.nan is also float. If you want to replace non-null values with the index values, you can use df.where:

output = df.where(df.isna(), df.index.tolist())

Output:

           1     2
0       
index1  index1  index1
index2  index2  NaN
index3  NaN index3
index4  index4  index4