I want to replace the elements of an array (which contains numbers) with strings.For example I want "1" to be replaced with "a", I want "2" to be replaced with "b", and I want "3" to be replaced with "c". I did the following and didn't work for me. Can anyone suggest a way to resolve the issue?
test = np.where(test==1, "a", test)
test = np.where(test==2, "b", test)
test = np.where(test==3, "c", test)
your looking for np.select
it works by giving np.select a list of conditions, and a list of choices, and if the condition is true, it will replace that value with the same index from the choice list, if not, then it will use the default value, the 3rd positional argument in np.select
CODE:
import numpy as np
test = np.array([1,2,3,4,5,6,7,1,3,2,5])
condlist = [test==1 ,test==2, test==3]
choicelist = ['a', 'b', 'c']
test = np.select(condlist, choicelist, test)
test
OUTPUT:
array(['a', 'b', 'c', '4', '5', '6', '7', 'a', 'c', 'b', '5'],
dtype='<U11')
In your code, after the first modification, dtype of the array will be changed from int or float to <U11, so, test==2 and test==3for the next lines must be written as strings. If:
test = np.array([[1, 5, 6], [1, 3, 2]])
# dtype --> int32
So, after the first line:
test = np.where(test==1, "a", test)
# dtype --> <U11
So, 2 and 3 must be written as string:
test = np.where(test=='2', "b", test)
test = np.where(test=='3', "c", test)
It is the cause of the error. np.select is another way.
numpy.ndarrayscannot contain mixed types, unless you plan to store plainobjects (defeating most of the reasons to usenumpy) or numbers as strings.testarraydtypeat each step.