NumPy | choose method
Start your free 7-days trial now!
Numpy's choose(~) method constructs a new array from a subset of the input array. The way that the subset is chosen is quite unique, and explaining via examples would be far better than via words, so please check out the examples below!
Parameters
1. a | array-like of int
An array of integer indices you want to extract.
2. choices | sequence of arrays
The superset.
3. out | Numpy array | optional
A Numpy array to place the extracted subset.
4. mode | string | optional
How to deal with indices that are out of bounds:
Mode | Description |
|---|---|
raise | Throw an error. |
wrap | Go through another cycle around the array. |
clip | Get the last element of the array. |
By default, mode="raise".
Return value
A Numpy array containing the specified subset.
Examples
1D arrays
To extract the 1st and 3rd index from a 1D array:
a = np.array([4,5,6,7])np.choose([1,3], a)
array([5, 7])
2D arrays
a = np.array([[4,5,6],[7,8,9],[10,11,12]])np.choose([1,0,1], a)
array([ 7, 5, 9])
Here, we are doing the following in order:
extracting the value at the 0th index of the 1+1=2nd array (i.e. [7,8,9]), which is 7.
extracting the value at the 1st index of the 0+1=1st array (i.e. [4,5,6]), which is 5.
extracting the value at the 2nd index of the 1+1=2nd array (i.e. [7,8,9]), which is 9.
The first argument, [1,0,1], just means that we want to extract a value from array at indexes 1, 0 and 1. The specific value that will be chosen will depend on the ordering:
the first element in the resulting array will be the 0th index of the array located at index 1.
the second element in the resulting array will be the 1st index of the array located at index 0.
and so on.
Here's yet another example to test your understanding:
a = np.array([[4,5,6],[7,8,9],[10,11,12]])np.choose([2,0,1], a)
array([10, 5, 9])
Different modes
raise
The default parameter value for mode is raise:
a = np.array([7,8,9])np.choose([4], a, mode="raise")
ValueError: invalid entry in choice array
Here, we get an error because the index we specified (4) is out of bounds.
wrap
a = np.array([7,8,9])np.choose([4], a, mode="wrap")
array([8])
Here, since the index 4 does not exist, we go through another cycle around the array; 4-3=1st index is then selected.
clip
a = np.array([7,8,9])np.choose([4], a, mode="clip")
array([9])
Here, since index 4 is out of bounds, we just get the last element of the array, 4.
