How could I write a numpy function where it only filters out the array strings that ends with 'USD'. How would I be able to execute this filter without a for loop.
import numpy as np
Array= ['BTCUSD', 'ETHUSD', 'David', 'georGe', 'XRPUSD', 'USDAUD', 'ETHUSDC' ]
Expected Output
['BTCUSD', 'ETHUSD', 'XRPUSD']
Using numpy char.endswith
import numpy as np
a = np.array(['BTCUSD', 'ETHUSD', 'David', 'georGe', 'XRPUSD', 'USDAUD', 'ETHUSDC'])
print(a[np.char.endswith(a, 'USD')])
Output:
['BTCUSD' 'ETHUSD' 'XRPUSD']
For a return type of list instead of np.ndarray a comprehension can be used:
import numpy as np
lst = np.array(['BTCUSD', 'ETHUSD', 'David', 'georGe', 'XRPUSD', 'USDAUD', 'ETHUSDC'])
print([elem for elem in lst if elem.endswith('USD')])
Output:
['BTCUSD', 'ETHUSD', 'XRPUSD']
*The comprehension approach can be used on Python lists as well as np arrays.
np.char functions have about the same speed as list comprehensions. They still use python string methods.
a[np.char.endswith(a, 'USD', 'USDT')] however doesnt really workAbout three years later, the accepted answer is now deprecated. The np.char family of functions only operated on fixed length string data and was superseded by the new np.string functions.
The old functions from np.char and the ones in numpy.char.chararray are planned to be removed in a future release of NumPy.
When using NumPy from version 2.0 or newer, the functions from np.string should be preferred. See: https://numpy.org/doc/stable/reference/routines.char.html#module-numpy.char
Currently, the accepted way of doing what was requested by the OP would be using the np.string.endswith:
>>> import numpy as np
>>> a = np.array(['BTCUSD', 'ETHUSD', 'David', 'georGe', 'XRPUSD', 'USDAUD', 'ETHUSDC'])
>>> a[np.strings.endswith(a, 'USD')]
array(['BTCUSD', 'ETHUSD', 'XRPUSD'], dtype='<U7')
numpydoesn't have fast string methods, so beware of trying to replace list/string methods withnumpy.