How does Series convert from object to number?

I have a pandas.series which consist zip_codes like as: '23123213', '212212', '54545' and some zip_codes looks like: '4344-4343', '7676432-878'. I create def for changing each elements of series to numeric format:

def numbers_zip_code(a): # function for changing type of zip_code from object to number
try:
    a = int(a)
except ValueError:
    splitted_a = a.split('-')
    a = int(splitted_a[0]) + int(splitted_a[1])
return a

I used this function:

X.zip_code = X.zip_code.map(lambda x: numbers_zip_code(x))

After that I could use for this series method describe():

X.zip_code.describe()

Results:

count 159880
unique 3498
top 48227
freq 7316
Name: zip_code, dtype: int64

Also I checked each element in series using circle for:

object_list = [] #create lists with columns type object and another with numbers
number_list = []
for i in X.zip_code:
    if i == 'object':
        object_list.append(i)
    else:
        number_list.append(i)
len(object_list)

Result: 0

And it's fine... but when I check whole dataframe (series with zip_codes is a part of df) and for this column (zip_code) type again is 'object' !!!! Really, I don't understand I should do...

X.dtypes

Results:

ticket_id int64
agency_name object
inspector_name object
violator_name object
violation_street_number float64
violation_street_name object
mailing_address_str_number float64
mailing_address_str_name object
city object
state object
zip_code object
etc

Solution

Try:

X['zip_code'] = X['zip_code'].astype('int64')