Python: assign boolean values to list elements given a certain condition

Question

Say you have the following list:

import numpy as np

mylist=[8.4468975,
        8.52036,
        9.3605785,
        np.ma.core.MaskedConstant(),
        np.ma.core.MaskedConstant()]  

 # [8.4468975, 8.52036, 9.3605785, masked, masked]

and you want to assign 0 to all values that are masked, and 1 to all the others. How can you properly do it?

I've tried this, but the result is an empty list:

mylist=[1 for i,v in enumerate(mylist) if type(mylist[i]) is float]

The result should be numerical (no True/False):

mylist=[1,1,1,0,0]

EDIT

The masked values that are present in my original list are of the type numpy.ma.core.MaskedConstant.

Ma0 · Accepted Answer · 2017-03-15 09:10:39Z

2

Like this:

mylist=[8.4468975,
        8.52036,
        9.3605785,
        masked,
        masked]

mylist = [0 if type(x) is numpy.ma.core.MaskedConstant else 1 for x in mylist]

Also note that your list comprehension mylist=[1 for i,v in enumerate(mylist) if type(mylist[i]) is float] does not quite do what you describe; at least not in the general case. You are assigning 1 to anything that is not float which is not the same as to anything that is 'masked'.

Also note that when looping like [1 for i,v in enumerate(mylist) if type(mylist[i]) is float] you are (falsely though, see further down) not using the v so you could rewrite as [1 for i,_ in enumerate(mylist) if type(mylist[i]) is float]. Notice the underscore. It is used to imply that you will be dumping the second item of the tuple returned by enumerate().

Finally, i said falsely further up because in your list comprehension mylist[i] can be substituted by v. So properly done it would be written like: [1 for v in mylist if type(v) is float]. But as I said, this is slightly different from what you seem to want.

edited Mar 15, 2017 at 9:10

answered Mar 15, 2017 at 8:56

Ma0

15.2k4 gold badges38 silver badges70 bronze badges

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

2 Comments

FaCoffee Over a year ago

I had actually forgot an important thing regarding the type of the masked instances. They are numpy.ma.core.MaskedConstant, and not strings. So your approach does not work the way it is now.

Ma0 Over a year ago

@CF84 Now your workaround makes more sense. I will read a bit on those and edit my answer

Chris_Rands · Accepted Answer · 2017-03-15 09:20:53Z

2

You can use int() to cast your booleans to integers:

[int(isinstance(x,float)) for x in mylist]

Note this works because numpy.ma.core.MaskedConstant() is not a float.

edited Mar 15, 2017 at 9:20

answered Mar 15, 2017 at 9:10

Chris_Rands

41.7k15 gold badges92 silver badges126 bronze badges

Comments

FLab · Accepted Answer · 2017-03-15 09:08:27Z

1

You can try this:

mylist = [1 if isinstance(v, float) else 0 for i,v in enumerate(mylist)]

Alternatively, you could do:

mylist = [0 if isinstance(v, np.ma.core.MaskedConstant) else 1 for i,v in enumerate(mylist)]

so that you only set to 0 a specific mask type.

The relevant function to check for the type of a variable is: https://docs.python.org/2/library/functions.html#isinstance

Moreover, as you (correctly) use the enumerate construct, you already looping over the values with variable v, so no need to access it doing mylist[i].

edited Mar 15, 2017 at 9:08

answered Mar 15, 2017 at 8:59

FLab

7,5465 gold badges40 silver badges70 bronze badges

6 Comments

FaCoffee Over a year ago

I had actually forgot an important thing regarding the type of the masked instances. They are numpy.ma.core.MaskedConstant, and not strings. So your approach does not work the way it is now.

FLab Over a year ago

My approach should work anyway, since it does not rely on string comparison.

FaCoffee Over a year ago

The result I get is a list of 0 values. Nothing else. I don't see why

FLab Over a year ago

Please try the second approach

FLab Over a year ago

simple typo, was supposed to be "v". Fixed now

|

akash karothiya · Accepted Answer · 2017-03-15 09:12:25Z

1

You can use map function here :

>>> list(map(lambda x : 0 if isinstance(x,np.ma.core.MaskedConstant) else 1 , mylist))

[1,1,1,0,0]

answered Mar 15, 2017 at 9:12

akash karothiya

5,9601 gold badge21 silver badges29 bronze badges

Comments

Elmex80s · Accepted Answer · 2017-03-15 09:19:16Z

1

You do not see this answer that often therefore I give it despite the other answers are satisfactory. Define a function f

def f(x):
    if isinstance(x, np.ma.core.MaskedConstant):
        return 0
    else:
        return 1

Now do list comprehension

mylist = [f(x) for x in mylist]

edited Mar 15, 2017 at 9:19

answered Mar 15, 2017 at 9:13

Elmex80s

3,5041 gold badge17 silver badges25 bronze badges

2 Comments

FaCoffee Over a year ago

Thanks, this is really helpful as it puts things into perspective and reminds me of the existence of functions. However, given the lengthy code I am working on, I find it more suitable to follow a list-comprehension approach.

Elmex80s Over a year ago

@CF84 This is the list comprehension approach.

Collectives™ on Stack Overflow

Python: assign boolean values to list elements given a certain condition

5 Answers 5

2 Comments

Comments

6 Comments

Comments

2 Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

5 Answers 5

2 Comments

Comments

6 Comments

Comments

2 Comments

Your Answer

Sign up or log in

Post as a guest

Related