Changing Column Type in Pandas DataFrame to int64

Question

I am trying to change a column's data type from type: object to type: int64 within a DataFrame using .map().

   df['one'] = df['one'].map(convert_to_int_with_error)

Here is my function:

def convert_to_int_with_error(x):
    if not x in ['', None, ' ']:
        try:
            return np.int64(x)
        except ValueError as e:
            print(e)
            return None
    else:
        return None

    if not type(x) == np.int64():
        print("Not int64")
        sys.exit()

This completes successfully. However, when I check the data type after completion, it reverts to type: float:

print("%s is a %s after converting" % (key, df['one'].dtype))

Where exactly did you put the if not type(x) == np.int64(): condition? Are you saying that convert_to_int_with_error never returns None? — unutbu
– unutbu, Commented Dec 30, 2016 at 14:47
For numerical containers, None will be regarded as NaN so as to keep it's float(numerical) dtype. You need to find a way to handle such missing values/empty strings so that it would result in np.int64 dtype. — Nickil Maveli
– Nickil Maveli, Commented Dec 30, 2016 at 14:53

jezrael · Accepted Answer · 2016-12-30 14:43:36Z

2

I think problem is your problematic values are converted from None to NaN, so int is cast to float - see docs.

Instead map you can use to_numeric with parameter errors='coerce' for convert problematic values to NaN:

df['one'] = pd.to_numeric(df['one'], errors='coerce')

answered Dec 30, 2016 at 14:43

jezrael

868k103 gold badges1.4k silver badges1.3k bronze badges

Sign up to request clarification or add additional context in comments.

2 Comments

sccrthlt Over a year ago

I included the try and except to account for values that could not be converted to int64 properly though?

jezrael Over a year ago

Unfortunately is not possible have dtype int with NaN or None values.

Collectives™ on Stack Overflow

Changing Column Type in Pandas DataFrame to int64

1 Answer 1

2 Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

2 Comments

Your Answer

Sign up or log in

Post as a guest

Related