How to Efficiently Select Specific Columns per Row in NumPy Using Lists or Boolean Arrays?

Efficiently Selecting Specific Columns per Row in NumPy Using Lists or Boolean Arrays

NumPy offers extensive capabilities for manipulating multidimensional arrays. However, selecting specific columns based on a list of indexes for each row can be challenging and demand efficient solutions.

One approach to solving this issue is to utilize boolean arrays. Each column of a boolean array can represent the desired selection for a particular row. By using direct selection with the boolean array, specific columns can be extracted efficiently. For instance:

<code class="python">import numpy as np

a = np.array([1, 2, 3])
b = np.array([[False, True, False], [True, False, False], [False, False, True]])
a[b]
# Output: [2, 4, 9]</code>

Alternatively, it's possible to create an array representing the range of columns and use direct selection on it. This approach, however, may not always be optimal.

<code class="python">a = np.array([[1, 2, 3], [4, 5, 6], [7, 8, 9]])
a[np.arange(len(a)), [1, 0, 2]]
# Output: [2, 4, 9]</code>

By leveraging these methods, it's feasible to efficiently select specific columns per row in NumPy arrays, regardless of whether the selection criteria are provided as a list of indexes or a boolean array.

The above is the detailed content of How to Efficiently Select Specific Columns per Row in NumPy Using Lists or Boolean Arrays?. For more information, please follow other related articles on the PHP Chinese website!