Multiple dtypes in a Numpy array

Question

I have the following data set in a numpy array:

Array 1:

[[a, 1, 20]
 [a, 3, 40]
 [b, 1, 20]
 [b, 2, 40]
 [c, 5, 90]]

Array 2:

[[a, 2]
 [a, 5]]

What I'm trying to accomplish is the following: Array2[0,0]=a, and Array2[0,1]=2 I want to interpolate from the first array to find a,2,30.

To do this I'm using np.where(Array1==item)[0] which looks for 'a', I can't interpolate though because the dtype used to import was a string, not an int.

It's been a while since I've used Numpy so if I'm completely in the weeds please let me know.

What does it mean to interpolate between 'a' and 2? And what other dtype would you use for 'a'? — abarnert
– abarnert, Commented Dec 6, 2014 at 0:46
What is a. Is it a string? a variable? What's the dtype of these arrays? — hpaulj
– hpaulj, Commented Dec 6, 2014 at 2:08

abarnert · Accepted Answer · 2014-12-06 00:59:22Z

1

I'm not entirely clear on what you're trying to do, but it sounds like you want to specify an aggregate dtype.

This is explained in detail in the dtype docs.

For example, here's a way to specify that each row has a 1-character string and a 64-bit native float (when you don't care what the field names are):

dt = np.dtype('U1, f8')

There are of course other ways to write this; read the full page for details.

And, assuming you've read this in with loadtxt, the docs there have a nice example of using such a dtype. For example:

>>> s2 = 'a 2\na 5\n'
>>> i2 = io.StringIO(s2)
>>> a2 = np.loadtxt(i2, 'U1, i4')
>>> a2
array([('a', 2), ('a', 5)],
      dtype=[('f0', '<U1'), ('f1', '<i4')])

edited Dec 6, 2014 at 0:59

answered Dec 6, 2014 at 0:54

abarnert

368k54 gold badges626 silver badges691 bronze badges

Sign up to request clarification or add additional context in comments.

1 Comment

VectorVictor Over a year ago

dt = np.dtype('U1, f8') did not work for me? this behaved like a float. I could not find any reference in the docs

Collectives™ on Stack Overflow

Multiple dtypes in a Numpy array

1 Answer 1

1 Comment

Your Answer

Linked

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

1 Comment

Your Answer

Sign up or log in

Post as a guest

Linked

Related