Dropping rows from pandas dataframe with some function on the series

Question

I am trying to drop outdated values from a dataframe that holds a difficult to compare date format. My code:

frame=pd.read_csv("somefile.csv")
frame=frame[datetime.strptime(frame["CreationTime"],"%d-%b-%Y %H:%M:%S")>(datetime.now()-timedelta(days=daypar))]

However I get the error that the 1st arg of strptime should be str and not Series, Which makes sense. So how would one go about solving this issue?

You can use `pd.to_datetime(frame['CreationTime'], format="%d-%b-%Y %H:%M:%S")` to convert to datetime and drop in a second step. — Dschoni, Mar 11 '21 at 12:37
Also see this answer for parsing dates directly on loading: https://stackoverflow.com/questions/17465045/can-pandas-automatically-recognize-dates — Dschoni, Mar 11 '21 at 12:39
I tried ```pd.to_datetime(frame['CreationTime'], format="%d-%b-%Y %H:%M:%S")``` which errored with " TypeError: Unrecognized value type: " — MegaFlipFlop, Mar 11 '21 at 13:03

score 1 · Accepted Answer · answered Mar 11 '21 at 13:02

1

Pandas read_csv is clever enough to decode timestamps at load time:

frame=pd.read_csv("somefile.csv", parse_dates=['CreationTime'])
frame=frame[df['CreationTime']>(datetime.now()-timedelta(days=daypar))]

answered Mar 11 '21 at 13:02

Serge Ballesta

1 Answers1