Clever Geek Handbook

How to round up datetime64 values

I have a numpy array from datetime64

, and I would like to round the sub-subset values ​​of the array elements. For example, from 2001-1-1 10:33:32.5

to 2001-1-1 10:33:32.0

. I am looking for a vecotrized method.

More generally, I am looking for a vectorial method to round to any frequency (minutes, days, etc.).

+4


source to share


2 answers


rounded = numpy.array(myarray, dtype='datetime64[s]')

or rounded = myarray.astype('datetime64[s]')

This also works within minutes using:



rounded = numpy.array(myarray, dtype='datetime64[m]')
+6


source


Another method to handle truncation would be:



rounded = numpy.array((numpy.array(myarray, dtype='datetime64[m]'), dtype='datetime64[ms]')
0


source






More articles:


All Articles