Cannot convert the series to <class 'int'> when trying to create new dataframe column

[Mark17]

Hi all,

I have a dataframe df2 with a 'DATE' column that starts with 17169 in row 2. I'm trying to create a new column with a human readable format:

This...

df2['Human Date'] = datetime.utcfromtimestamp(17169*86400).strftime('%Y-%m-%d')

...returns '2017-01-03'. Why doesn't this...

df2['Human Date'] = datetime.utcfromtimestamp(df2['DATE']*86400).strftime('%Y-%m-%d')

...create a new column and populate based on the same row of the 'DATE' column? Instead, it gives the TypeError shown in the thread subject.

[Mark17]

Can anyone answer this? Thanks!

[Mark17]

This works:

df2['Human Date'] = pd.to_datetime(df2['DATE'],unit='D')

This also works:

df2.insert(1,'Human Date',pd.to_datetime(df2['DATE'],unit='D'))

I'd still like to know why my initial attempt led to the TypeError. Thanks!

[deanhystad]

datetime.utcfromtimestamp() expects a float, not a Series. I don't know why the message complains about<class 'int'>.

What you want to do is apply the datetime.utcfromtimestamp() function to each item in the series. For that you use "apply".

import pandas as pd
from datetime import datetime

df2 = pd.DataFrame({"DATE":[17169, 17200]})
df2['DATESTR'] = df2['DATE'].apply(lambda t: datetime.utcfromtimestamp(t*86400).strftime('%Y-%m-%d'))
print(df2)

Output:
    DATE     DATESTR
0  17169  2017-01-03
1  17200  2017-02-03