Numpy sorting by increasing datetime

Question

I have an first array i containing datetime elements (with the shape (33416,)) and a second array m containing floats values (with the shape (33416,20)). This correspong to 20 measurements made at a specific instant. I would like to sort i array according to increasing datetime values in i and sort lines of m array accordingly in order to have increasing instant of measurements and its corresponding 20 measurements at this instant. Can some help me ? or understand my needs ???

Answer 1

You can also use numpy for this. Assuming you have:

dates = numpy.array([datetime(2012,02,03,12,00,00), datetime(2012,02,03,15,00,00), datetime(2012,02,03,13,00,00)])
values = numpy.array([[1, 1], [3, 3], [2, 2]])

You can do at first:

unified = numpy.concatenate((dates.reshape(len(dates), 1), values), axis=1)

This will merge your two list, which considering what you want should be a better suited way to hold your data anyway. So now you have:

unified = array([[2012-02-03 12:00:00, 1, 1],
                 [2012-02-03 15:00:00, 3, 3],
                 [2012-02-03 13:00:00, 2, 2]], dtype=object)

Now you can do:

unified = numpy.sort(unified, axis=0)

This will sort after date. Now if you still want just the separate lists you can get them with:

unified[:, 0]

>>> array([2012-02-03 12:00:00, 2012-02-03 13:00:00, 2012-02-03 15:00:00], dtype=object)

unified[:, 1:]

>>> array([[1, 1],
           [2, 2],
           [3, 3]], dtype=object)

EDIT after your comment Ok now that I fully understand what you want you can achieve that by replacing

   unified = numpy.sort(unified, axis=0)    

with:

   unified = numpy.array(sorted(unified, key= lambda x: x[0]))

EDIT

have you tried what I just suggested? In my terminal:

    unified = numpy.array([[datetime(2012,02,03,12,00,00), 4, 1],[datetime(2012,02,03,15,00,00), 5, 2],[datetime(2012,02,03,13,00,00), 2, 1]], dtype=object)
    >>> unified
        array([[2012-02-03 12:00:00, 4, 1],
               [2012-02-03 15:00:00, 5, 2],
               [2012-02-03 13:00:00, 2, 1]], dtype=object)

    >>> unified = numpy.array(sorted(unified, key=lambda x: x[0]))
    >>> unified
        array([[2012-02-03 12:00:00, 4, 1],
               [2012-02-03 13:00:00, 2, 1],
               [2012-02-03 15:00:00, 5, 2]], dtype=object)