I´m trying to convert from pandas._libs.tslibs.timestamps.Timestamp to datetime.datetime but the change is not saved:
type(df_unix.loc[45,'LastLoginDate'])
OUTPUT: pandas._libs.tslibs.timestamps.Timestamp
type(df_unix.loc[45,'LastLoginDate'].to_pydatetime())
OUTPUT: datetime.datetime
df_unix.loc[45,'LastLoginDate'] = df_unix.loc[i,'LastLoginDate'].to_pydatetime()
type(df_unix.loc[45,'LastLoginDate'])
OUTPUT: pandas._libs.tslibs.timestamps.Timestamp
Suppose that you have this date:
some_date = pd.Timestamp("1970-01-01")
print(some_date)
Output: Timestamp('1970-01-01 00:00:00')
To transform it to datetime just do this:
some_date.to_pydatetime()
And you will get:
output: datetime.datetime(1970, 1, 1, 0, 0)
You can't use datetime.datetime(some_date) because the constructor of the datetime.datetime class takes 3 int numbers:
One for the year, one for the month and one for the day.
df_unix.loc[45,'LastLoginDate'] returns a scalar value. pandas uses Timestamp for scalar values so the type is correct.
Run df_unix.info(). You should see the column type is still datetime64[ns].
You are also assigning one value at location 45 using to_pydatetime which is used for returning the native Python datetime object. It would be an interesting result if this changed the type for the whole column.
