I have a regular float ndarray with shape (100,50). How can I give names to the first two columns, i.e. 'idx1' and 'idx2', and 'data' to the rest of the columns.
Add a comment
|
1 Answer
You can use the view method. Here's an example, using an array with shape (3, 5) for the demo:
In [21]: x
Out[21]:
array([[ 0., 1., 2., 3., 4.],
[ 5., 6., 7., 8., 9.],
[ 10., 11., 12., 13., 14.]])
In [22]: y = x.ravel().view(dtype=[('idx1', float), ('idx2', float), ('data', float, 3)])
In [23]: y['idx1']
Out[23]: array([ 0., 5., 10.])
In [24]: y['data']
Out[24]:
array([[ 2., 3., 4.],
[ 7., 8., 9.],
[ 12., 13., 14.]])
In [25]: y['data'][1]
Out[25]: array([ 7., 8., 9.])
Note that y is a 1-D array; it has shape (3,). If you change the conversion to y = x.view(...) (i.e. don't ravel x), y will have have (3,1), and y['idx1'] will have shape (3,1).
1 Comment
askewchan
Nice, didn't know about using
ravel to avoid the extra length-1 axis.