How to modify a Numpy array with conditional and function

Question

I have a 3d with about 8e6 elements, and i need to change all of the elements. I want to do something like this

def func(x, index):
....


flat = array.reshape(-1)
flat[flat == 9999] = 0
flat[flat >= 0.2 and flat < 0.7] = func(flat, 0)
flat[flat >= 0.7 and flat < 1.5] = func(flat, 1)
flat[flat >= 0.7 and flat < 1.5] = func(flat, 2)
....
...

This code doesnt work. I've tried np.nditer but it wouldnt let me retrieve the value per index, it seems to be retrieving the whole array. in short of looping w/ a for loop per dimension and read and write each value, is there another way i can do this?

Thanks

Imanol Luengo · Accepted Answer · 2015-07-08 14:46:03Z

4

While using multiple conditionals for masking, you need to use bitwise operator &:

flat[(flat >= 0.2) & (flat < 0.7)] = ...

Additionally, as your mask dimensions and the result of the func might not be the same (looks to me that func does apply something to the whole array), you should do something like:

mask = (flat >= 0.2) & (flat < 0.7)
flat[mask] = func(flat, 0)[mask]

Or like this if your func is elementwise:

mask = (flat >= 0.2) & (flat < 0.7)
flat[mask] = func(flat[mask], 0)

Then repeat the same process for the other 2 conditions.

As a bonus, you can iterate over all your lower bounds ranges and do like this:

lower_bound = [0.2, 0.7, 1.5]
upper_bound = lower_bound[1:] + [np.inf]

for i in range(len(lower_bound)):
    mask = (flat >= lower_bound[i]) & (flat < upper_bound[i])
    flat[mask] = func(flat[mask], i)

That would avoid you of writing 100 lines if you have 100 ranges.

edited Jul 8, 2015 at 14:46

answered Jul 8, 2015 at 14:29

Imanol Luengo

16k3 gold badges52 silver badges68 bronze badges

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

1 Comment

Vinh Over a year ago

thanks that works. as of now i would need 10 masks is there a better way than creating 10 masks?

YXD · Accepted Answer · 2015-07-08 14:42:42Z

2

The syntax is flat[(flat >= 0.2) & (flat < 0.7)] and similar. Note the right hand side of flat[--] = func(--)has to have the same number of elements, so you want call func on the indexed/masked array.

There are some NumPy functions that can help with this kind of code:

answered Jul 8, 2015 at 14:42

YXD

32.6k15 gold badges79 silver badges117 bronze badges

Collectives™ on Stack Overflow

How to modify a Numpy array with conditional and function

2 Answers 2

1 Comment

Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

2 Answers 2

1 Comment

Comments

Your Answer

Sign up or log in

Post as a guest

Related