Removing values with mixed datatypes

Question

I have the following DataFrame

A       B       C
1.0     abc     1.0
abc     1.0     abc
-1.11   abc     abc

I have mixed datatypes (float and str). How can I drop values <= -1 in column A.

I get an error if I do the following because of the mixed datatypes

df['A'] = (df['A'] != "abc") & (df['A'] > -1)
TypeError: '>' not supported between instances of 'str' and 'int'

How can I change my object to make abc a str and 1.0 a float so I can:

(df['A'] != "abc") & (df['A'] > -1)

print(df['A'].dtype)
    -> object

I would like the expected output

df = 

A       B       C
1.0     abc     1.0
abc     1.0     abc
NaN     abc     abc

Answer 1

There are at least a couple of different approaches to this problem.

loc + pd.to_numeric

pd.DataFrame.loc accepts Boolean series, so you can calculate a mask via pd.to_numeric and feed into the loc setter.

Note there is no need to specify df['A'] != 'abc' because the mask series will convert these values to NaN.

mask = pd.to_numeric(df['A'], errors='coerce') < -1
df.loc[mask, 'A'] = np.nan

print(df)

     A    B    C
0    1  abc    1
1  abc    1  abc
2  NaN  abc  abc

try / except

See @Jan's solution. This solution is preferable if you expect values to be numeric and are only looking for alternative treatment in edge cases.