Integer Array Indexing
Integer array indexing allows you to select specific elements using arrays of indices. This method differs significantly from slicing, as it creates a copy of the selected data rather than a view.
import numpy as np
arr = np.array([[1, 2, 3], [4, 5, 6], [7, 8, 9]])
print(arr[[0, 1, 2], [0, 2, 1]]) # Output: [1 6 8]
rows = np.array([0, 2])
cols = np.array([1, 2])
print(arr[rows, cols]) # Output: [2 9]
print(arr[1:, [0,2]]) # Output: [[4 6] [7 9]]This technique allows for flexible selection of non-contiguous elements, making it ideal for tasks like extracting specific patterns or features from data.
Boolean Array Indexing (Masking)
Boolean array indexing, or masking, is a powerful technique for selecting elements based on a condition. It creates a boolean array of the same shape as the input array, where True indicates selection and False indicates exclusion.
arr = np.array([10, 20, 30, 40, 50])
mask = arr > 25
print(arr[mask]) # Output: [30 40 50]
mask = (arr > 20) & (arr < 45)
print(arr[mask]) # Output: [30 40]
#Modifying array elements based on a boolean condition
arr[arr>30] = 100
print(arr) #Output: [ 10 20 30 100 100]Boolean indexing provides a concise and efficient way to filter data based on arbitrary criteria, crucial for data cleaning, filtering and analysis.
Mixing Integer and Boolean Indexing
NumPy allows you to combine integer and boolean indexing. Integer indexing is applied first, followed by boolean indexing on the result.
arr = np.array([[1, 2, 3], [4, 5, 6], [7, 8, 9]])
row = arr[1]
#Apply boolean indexing to filter elements greater than 4
print(row[row > 4]) # Output: [5 6]
#Combining it all together
print(arr[ [0,2], arr[ [0,2] ] >2 ]) #Output: [3 9]This combined approach offers fine-grained control over data selection, enabling sophisticated data manipulation within NumPy arrays.
Advanced Indexing with Multiple Dimensions
Advanced indexing extends seamlessly to multi-dimensional arrays. You can use integer or boolean arrays for each dimension to select specific subsets.
arr = np.arange(27).reshape((3, 3, 3))
#Selecting specific elements across multiple dimensions.
print(arr[[0, 2], [0, 1], [1, 2]]) #Output: [ 1 26]This allows for complex selections across multiple axes of your array making your data analysis easier. Remember that when using advanced indexing you are getting a copy of the data and not a view.