I want to print a string depending on the values in a NumPy array, e.g. value 0 should lead to letter 'a'.
import numpy as np
arr = np.zeros((2,2))
arr[(0,0)] = 1
arr[(0,1)] = 2
printValues = {0:'a', 1:'b', 2:'c'}
print(np.array2string(arr, formatter={'str':lambda x: printValues[x]}))
Expected result:
[['b' 'c']
['a' 'a']]
Observed:
[[1. 2.]
[0. 0.]]
The keyword argument formatter for array2string needs the type of the elements of the array which you want to replace, not the type which you're converting to.
So, in your example, instead of str you should use float, since 0., 1. and 2. are floats.
If you want to make sure, every element of the array is definitely printed with your formatter, use all:
import numpy as np
arr = np.zeros((2, 2))
arr[(0, 0)] = 1
arr[(0, 1)] = 2
printValues = {0: 'a', 1: 'b', 2: 'c'}
print(np.array2string(arr, formatter={'all': lambda x: printValues[int(x)]}))
See the above linked documentation for more available types. Maybe float_kind is also a good idea for you.
IIUC vectorize
np.vectorize(printValues.get)(arr)
array([['b', 'c'],
['a', 'a']], dtype='<U1')