Pandas Can't avoid SettingWithCopyWarning

Question

I have a dataframe like this:

A     B     C

15   'ds'   '    0.000'
32   'ds'   '    1.000'
56   'ds'   '    2,700.000'
45   'gb'   '    7.000'

I want to change the values of column C to integers; so what i'm doing is something like this:

df.loc[:,'C'] = df.loc[:,'C'].apply(lambda x: int(float(x.strip().replace(',',''))))

This do the job, however, i get that annoying SettingWithCopyWarning. Why this appear if i'm using loc?

Answer 1

Pandas raises this warning in some cases with false positive (i.e. when, based on the order of assignments, you could be assigning to a copy, but in the current scenario aren't). This answer is helpful: How to deal with SettingWithCopyWarning in Pandas?

... but, personally, when I'm using .loc and still receiving the warning, I take the step offered in the above answer and disable the warning: pd.options.mode.chained_assignment = None

Answer 2

I'd use the following approach:

In [292]: df['C'] = pd.to_numeric(df['C'].str.strip().str.replace(',', ''), errors='coerce')

In [293]: df
Out[293]:
    A   B       C
0  15  ds     0.0
1  32  ds     1.0
2  56  ds  2700.0
3  45  gb     7.0

In [294]: df.dtypes
Out[294]:
A      int64
B     object
C    float64
dtype: object