Ensure that an array has all values encoded as 1 and -1, not 1 and 0

Question

There is a 1 dimensional numpy array named "target_y". My task is to ensure that "target_y" has all values encoded as 1 and -1, not 1 and 0 before performing logistic regression. Once I do it, I need to assign it to "prepared_y" and return it. Can I write something like:

if target_y in 1,-1:
    prepared_y = target_y

Why not replace the `0` values with `-1`? A quick search for replacing values in an numpy array will show how this can be done. — S3DEV, Jan 09 '21 at 21:28
Are you searching for [`all()`](https://docs.python.org/3/library/functions.html#all)? Both `1` and `-1` will be interpreted as `True`, so `all(target_y)` will return `True` if there's no `0` items. — Olvin Roght, Jan 09 '21 at 21:28
Is the goal to convert all 0's to -1's, or to return a boolean stating whether the array is valid or not? — bicarlsen, Jan 09 '21 at 21:29
yes, the goal is to check if there are any 0's and convert them to -1's. — Sandhya Indurkar, Jan 09 '21 at 21:31
Does this answer your question? [Replacing Numpy elements if condition is met](https://stackoverflow.com/questions/19766757/replacing-numpy-elements-if-condition-is-met) — S3DEV, Jan 09 '21 at 21:35

M Z · Answer 1 · 2021-01-09T21:35:08.160

0

Checkout np.all:

We can do some quick boolean arithmetic to check if all the values are 1 or -1:

if np.all((target_y == 1) + (target_y == -1)): # checks if the array has all 1s and -1s
    prepared_y = target_y

If you want to change all 0s to -1s, slice for all 0s:

target_y[np.argwhere(target_y == 0)] = -1

edited Jan 09 '21 at 21:35

answered Jan 09 '21 at 21:31

M Z

4,571
2
13
27

Alternatively you could use `np.all(target_y)`. – Olvin Roght Jan 09 '21 at 21:35
@OlvinRoght Not *quite*, but close. `np.all(arr)` will be true for all non-zero arrays. It does not check to see if it is all -1 and 1 – M Z Jan 09 '21 at 21:36
Yes, I know, but it fits as a solution for this particular problem – Olvin Roght Jan 09 '21 at 21:37

bicarlsen · Answer 2 · 2021-01-09T21:45:50.183

0

For a List you can do:

def convert( l ):
  return [ -1 if e == 0 else e for e in l ]

For a numpy array

def convert( l ):
  return l[ l == 0 ] = -1

edited Jan 09 '21 at 21:45

answered Jan 09 '21 at 21:38

bicarlsen

1,241
10
27

1

`[e or -1 for e in l]` – Olvin Roght Jan 09 '21 at 21:39
The reason I did it this way is in case there are other values than just `1` and `0`, but indeed, if you can ensure that those are the only two values @OlvinRoght's suggestion will work, defaulting to -1 if e evaluates to False. – bicarlsen Jan 09 '21 at 21:42

Nitish · Answer 3 · 2021-01-16T19:43:01.437

0

One value can be checked by if statement, such as:

if 0 in test_list:
    return test_list

To be exact to your answer, you need a loop to check each value:

def list_checker(test_list):
    for i, num in enumerate(test_list):
        if num == 0:
            test_list[i] = -1
    return test_list

edited Jan 16 '21 at 19:43

answered Jan 09 '21 at 21:42

Nitish

392
2
7

score 0 · Accepted Answer · answered Jan 09 '21 at 21:44

0

To replace all 0 values with -1 in a numpy.array it’s as simple as:

a[a == 0] = -1

... where a is the 1 dimensional array.

answered Jan 09 '21 at 21:44

S3DEV

8,768
3
31
42

3

Thank you! This is the easiest solution to my question. – Sandhya Indurkar Jan 09 '21 at 21:53
My pleasure. Glad it was helpful. – S3DEV Jan 09 '21 at 22:17

Ensure that an array has all values encoded as 1 and -1, not 1 and 0

4 Answers4