CODEWITHSUNDEEP
pythonpython-3.xpandas
utks009
utks009

Reputation: 573

How to change column data type based on row value condition using pandas

I want to change the data type of column value based on row condition, but not working

Sample Data:
    eName   Value1  Value2  NValue
    sample1 XYZ XYZ XYZ
    sample2 102 XYZ 102 XYZ 102 XYZ
    sample3 3   3.000   
    sample4 3   3.000

I have tried this approach, Is there different approach I have to use.

data.loc[data.eName == 'sample3', ['Value1', 'Value2']].apply(pd.to_numeric)
data.loc[data.eName == 'sample4', ['Value1', 'Value2']].apply(pd.to_numeric)


Output as:
    eName   Value1  Value2  NValue
    sample1 XYZ     XYZ     XYZ
    sample2 102 XYZ 102 XYZ 102 XYZ
    sample3 3.00    3.00    
    sample4 3.00    3.00

Upvotes: 1

Views: 307

Answers (1)

jezrael
jezrael

Reputation: 862581

You can assign mask created by Series.isin to both sides of filtered DataFrame and if necessary add errors='coerce' for convert non numeric values to NaN:

m = data.eName.isin(['sample3','sample4'])
cols = ['Value1', 'Value2']
#if need all columns without eName 
#cols = df.columns.difference(['eName'])
data.loc[m, cols] = data.loc[m , cols].apply(pd.to_numeric, errors='coerce')
print (data)
     eName   Value1   Value2   NValue
0  sample1      XYZ      XYZ      XYZ
1  sample2  102 XYZ  102 XYZ  102 XYZ
2  sample3        3        3      NaN
3  sample4        3        3      NaN

If need all columns without first:

m = data.eName.isin(['sample3','sample4']).values
data.iloc[m, 1:] = data.iloc[m , 1:].apply(pd.to_numeric, errors='coerce')

Upvotes: 3

Related Questions