I want to create a new column that returns a value of 1 if the below condition is true and 2 if false and am not sure why the below isn't working?
t1 = x['timestamp_1'] < x['timestamp_2']
x['new'] = np.select([t1], [1], default=2)
3 Answers
use numpy where
#convert both columns to pd.to_datetime
x[['timestamp_1','timestamp_2']] = x[['timestamp_1','timestamp_2']].apply(pd.to_datetime,errors='coerce')
t1 = x['timestamp_1'] < x['timestamp_2']
x['new'] = np.where(t1,1,2)
which functions like:
np.where(cond,valueiftrue,valueiffalse)
3 Comments
Chris90
when I try to create a new column as x['new'] = np.where(t1,1,2) I get error as below ValueError: Length of values does not match length of index
Chris90
wait sorry - it worked but it is giving a value of 2 for all rows, even when it should be 1? do I need to convert to datetime prior? @anky_91
anky
@Chris90 : check edited answer. :) Hope that works for you.
Use np.where:
#Convert to datetime
x['timestamp_1'] = pd.to_datetime(x['timestamp_1'])
x['timestamp_2'] = pd.to_datetime(x['timestamp_2'])
t1 = x['timestamp_1'] < x['timestamp_2']
x['new'] = np.where(t1, 1, 2)
If the condition is true it will return 1 else it will return 2.
2 Comments
Chris90
Thanks, but when I try this it gives a value of 2 for all rows even when should be 1? t1 = x['time_1'] < x['time_2'] x['new'] = np.where(t1, 1, 2)
Mohit Motwani
Are your columns time1 and time2 converted to datetime type? If they are strings, it will work as you said. Convert them using
x['time1'] = pd.to_datetime(x['time1'] and similary for the other columnYou can use the intuitive list comprehension after you convert your timestamps (pd.to_datetime), like this:
df['new'] = [1 if x==True else 2 for x in list(df.timestamp_1 < df.timestamp_2)]
Comments
np.where(cond,valueiftrue,valueiffalse)pd.x['']a typo? should it bex['column']?