How to make a numpy recarray with datatypes (datetime,float)?

numpy, python, recarray

Solution

You can use tuples instead of lists for the records:

>> y = [(datetime.datetime( 2012,1,1,1,1), 2.1), 
...     (datetime.datetime( 2012,1,1,1,2), -3.1),
...     (datetime.datetime( 2012,1,1,1,3), 0.1)]
>> np.rec.array(y, dtype=[('timestamp', object), ('x','f')])
rec.array([(datetime.datetime(2012, 1, 1, 1, 1), 2.0999999046325684),
           (datetime.datetime(2012, 1, 1, 1, 2), -3.0999999046325684),
           (datetime.datetime(2012, 1, 1, 1, 3), 0.10000000149011612)],
dtype=[('timestamp', '|O8'), ('x', '<f4')])

Problem

Let's say I have some simple data ``` y = [[datetime.datetime( 2012,1,1,1,1), 2.1], [datetime.datetime( 2012,1,1,1,2), -3.1], [datetime.datetime( 2012,1,1,1,3), 0.1]] ``` and I want a numpy record array corresponding to it. It would seem I ought to be able to do this: ``` np.rec.array( y, dtype=[('timestamp', object),('x','f')] ) ``` or this ``` np.rec.array( y, dtype=[('timestamp', '|O8'),('x','f')] ) ``` or maybe this ``` np.rec.array( y, dtype=[('timestamp', 'V'),('x','f')] ) ``` But each of them returns an error, either ``` ValueError: Setting void-array with object members using buffer. ``` or ``` TypeError: expected a readable buffer object ``` So how exactly can I set this up, assuming it is even possible?

Original source