Logical operations with array of strings in Python

Question

I know the following logical operation works with numpy:

A = np.array([True, False, True])
B = np.array([1.0, 2.0, 3.0])
C = A*B = array([1.0, 0.0, 3.0])

But the same isn't true if B is an array of strings. Is it possible to do the following:

A = np.array([True, False, True])
B = np.array(['eggs', 'milk', 'cheese'])
C = A*B = array(['eggs', '', 'cheese'])

That is a string multiplied with False should equal an empty string. Can this be done without a loop in Python (doesn't have to use numpy)?

Thanks!

Divakar · Accepted Answer · 2016-09-23 17:10:07Z

8

You can use np.where for making such selection based on a mask -

np.where(A,B,'')

Sample run -

In [4]: A
Out[4]: array([ True, False,  True], dtype=bool)

In [5]: B
Out[5]: 
array(['eggs', 'milk', 'cheese'], 
      dtype='|S6')

In [6]: np.where(A,B,'')
Out[6]: 
array(['eggs', '', 'cheese'], 
      dtype='|S6')

answered Sep 23, 2016 at 17:10

Divakar

222k19 gold badges273 silver badges374 bronze badges

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

Comments

hpaulj · Accepted Answer · 2016-09-23 18:38:32Z

3

np.char applies string methods to elements of an array:

In [301]: np.char.multiply(B, A.astype(int))
Out[301]: 
array(['eggs', '', 'cheese'], 
      dtype='<U6')

I had to convert the boolean to integer, and place it second.

Timing in other questions indicates that np.char iterates and applies the Python methods. Speed's about the same as for list comprehension.

For in-place change, use masked assignment instead of where

In [306]: B[~A]=''
In [307]: B
Out[307]: 
array(['eggs', '', 'cheese'], 
      dtype='<U6')

answered Sep 23, 2016 at 18:38

hpaulj

233k14 gold badges260 silver badges392 bronze badges

Comments

Łukasz Rogalski · Accepted Answer · 2016-09-23 17:11:35Z

2

Since strings may be multiplied by integers, and booleans are integers:

A = [True, False, True]
B = ['eggs', 'milk', 'cheese']
C = [a*b for a, b in zip(A, B)]
# C = ['eggs', '', 'cheese']

I still uses some kind of loop (same as numpy solution), but it's hidden in concise list comprehension.

Alternatively:

C = [a if b else '' for a, b in zip(A, B)]  # explicit loop may be clearer than multiply-sequence trick

answered Sep 23, 2016 at 17:11

Łukasz Rogalski

23.3k10 gold badges63 silver badges93 bronze badges

2 Comments

user2357112 Over a year ago

"I still uses some kind of loop (same as numpy solution), but it's hidden in concise list comprehension." - generally, when working with NumPy, you want your loops to be happening in C, not in list comprehensions. C loops get to avoid a ton of overhead. Python loops are generally somewhere from dozens to thousands of times slower.

Łukasz Rogalski Over a year ago

@user2357112 To be honest it's not clear to me whether OP uses numpy because he really needs powerful linear algebra toolbox or just because it's the only way he know to do piecewise operations. Storing strings in numpy arrays is pretty peculiar, it's not like you'll do matrix-vector multiplication with it... I just provided an alternative that doesn't have to use numpy. Feel free to up vote or down vote.

Collectives™ on Stack Overflow

Logical operations with array of strings in Python

3 Answers 3

Comments

Comments

2 Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

3 Answers 3

Comments

Comments

2 Comments

Your Answer

Sign up or log in

Post as a guest

Related