I'm making using a panda frame containing columns like below:
data = {'chindice': [ '-1', '5.89 e-06', '6.76 e-06', '6.31 e-06', '1',
'4', np.nan],
'target': ['classe1', 'classe2', 'classe3', np.nan,'classe5', 'classe4', 'classe5' ],
}
df = pd.DataFrame(data)
I need to use the columns "chindice" as float, but I couldnt because the columns dtype is 'object'. Any help would be appreciated. I am a newbie to pandas. Thanks
You can use to_numeric after stripping the problematic space in your scientific notation entries using str.replace:
In [15]:
df['chindice'] = pd.to_numeric(df['chindice'].str.replace(' ',''), errors='force')
df
Out[15]:
chindice target
0 -1.000000 classe1
1 0.000006 classe2
2 0.000007 classe3
3 0.000006 NaN
4 1.000000 classe5
5 4.000000 classe4
6 NaN classe5
Don't worry about the display, the real value is still there:
In [17]:
df['chindice'].iloc[1]
Out[17]:
5.8900000000000004e-06
You can use replace arbitrary whitespace \s+ and then cast by astype to float:
df['chindice'] = df.chindice.str.replace(r'\s+','').astype(float)
print df
chindice target
0 -1.000000 classe1
1 0.000006 classe2
2 0.000007 classe3
3 0.000006 NaN
4 1.000000 classe5
5 4.000000 classe4
6 NaN classe5
#temporaly display with precision 8
with pd.option_context('display.precision', 8):
print df
chindice target
0 -1.00000000 classe1
1 0.00000589 classe2
2 0.00000676 classe3
3 0.00000631 NaN
4 1.00000000 classe5
5 4.00000000 classe4
6 NaN classe5
