I have an array like:
array = ['A0','A1','A2','A3','A4','B0','B1','C0']
and want to obtain an array which is true for values with an A followed by a number ranging from 0 to 2.
So far, this is the way I do it:
selection = np.where ((array == 'A0') | (array == 'A1') | (array == 'A2'), 1, 0)
But is there a more elegant way to do this by using e.g., a regular expresion like:
selection = np.where (array == 'A[0-1]', 1, 0)
If using pandas is an option:
import numpy as np
import pandas as pd
a = np.array(['A0','A1','A2','A3','A4','B0','B1','C0'])
pd.Series(a).str.match(r'A[0-2]')
# 0 True
# 1 True
# 2 True
# 3 False
# 4 False
# 5 False
# 6 False
# 7 False
# dtype: bool
I don't think numpy if your best solution here. You can accomplish this using built-in python tools such as map.
import re
array = ['A0','A1','A2','A3','A4','B0','B1','C0']
p = r'A[0-2]'
list(map(lambda x: bool(re.match(p, x)), array))
# returns
[True, True, True, False, False, False, False, False]
# to get an array:
np.array(list(map(lambda x: bool(re.match(p, x)), array)))
# returns:
array([ True, True, True, False, False, False, False, False])
If it's not more complicated than A0, A1 and A2, you can use
a = np.array(['A0','A1','A2','A3','A4','B0','B1','C0'])
np.in1d(a, ['A0', 'A1', 'A2'])
# array([ True, True, True, False, False, False, False, False])
Try vectorization with blend of re :
import re
array = ['A0','A1','A2','A3','A4','B0','B1','C0']
y = np.vectorize(lambda y, x: bool(re.compile(x).match(y)))
selection = np.where(y(array, 'A[0-2]'), 1, 0)
print(selection)
#output:
[1 1 1 0 0 0 0 0]
bool(re.compile(x).match(y))
np.vectorize() is only for convenience, it doesn't provide any speedup over a for loop.
y confused meYou can also use list comprehension:
r = re.compile('A[0-2]')
selection = np.array([1 if re.match(r, i) else 0 for i in array])
You dont even need the lambda to vectorize a re.match():
array = ['A0','A1','A2','A3','A4','B0','B1','C0']
selection = np.vectorize(re.match,excluded={0})(r'A[0-1]', array)!= None)
its not any faster than the for loop, but its very simply. "excluded={0}" tells np.vectorize not to vectorize the first argument, the regex string.
