Numpy - delete data rows with negative values

Question

I am taking the output from AUTO numerical continuation package, and need to filter out results that have negative values of the variables, as they are non-physical. So if I have, for example:

>>>   a = np.array([[0,1,2,3,4],[-1,-0.5,0,0.5,1],[-3,-4,-5,0.1,0.2]])

I would like to be left with:

>>> b
array([[ 3. ,  4. ],
       [ 0.5,  1. ],
       [ 0.1,  0.2]])

But when I try numpy.where I get:

>>> b = a[:,(np.where(a[1]>=0) and np.where(a[2]>=0))]
>>> b
array([[[ 3. ,  4. ]],

       [[ 0.5,  1. ]],

       [[ 0.1,  0.2]]])
>>> b.shape
(3, 1, 2)

That is, it adds another unwanted axis to the array. What am I doing wrong?

You columns that are entirely positive, right ?

jrjc
– jrjc

2016-03-23 15:13:05 +00:00
Commented Mar 23, 2016 at 15:13 — jrjc
– jrjc, Commented Mar 23, 2016 at 15:13

hilberts_drinking_problem · Accepted Answer · 2016-03-23 15:13:46Z

8

Assuming all you want to do is to remove columns that have one or more negative values, you could do this:

a = np.array([[0,1,2,3,4],[-1,-0.5,0,0.5,1],[-3,-4,-5,0.1,0.2]])
b = a[:,a.min(axis=0)>=0]

answered Mar 23, 2016 at 15:13

hilberts_drinking_problem

11.6k3 gold badges25 silver badges55 bronze badges

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

Comments

Imanol Luengo · Accepted Answer · 2016-03-23 15:34:36Z

1

If what you want are fully positive columns, then @Yakym's answer is the way to go as it is probably faster. However, if it was just an example and you want to threshold certain columns, you can do it by sightly modifying your example:

>>> a[:, (a[1] >= 0) & (a[2] >= 0)]
array([[ 3. ,  4. ],
       [ 0.5,  1. ],
       [ 0.1,  0.2]])

Here (a[1] >= 0) and (a[2] >= 2) create boolean mask that are merged by the & (boolean/logical and) operator and used to index the array a.

answered Mar 23, 2016 at 15:34

Imanol Luengo

16k3 gold badges52 silver badges68 bronze badges

Collectives™ on Stack Overflow

Numpy - delete data rows with negative values

2 Answers 2

Comments

Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

2 Answers 2

Comments

Comments

Your Answer

Sign up or log in

Post as a guest

Related