Accessing and Modifying NumPy Arrays

In our last lesson, we introduced the concept of Python packages and NumPy in particular. Short for Numerical Python, this package is used to create and manipulate arrays and matrices that are essential to data science processes.

A critical aspect of working with NumPy is understanding how to efficiently access and manipulate the entries within an array. The ability to retrieve, modify, and analyze specific elements of an array isn’t just a technical necessity — it’s a gateway to more advanced data processing and analysis techniques. By leveraging the right methods for accessing array entries, you can optimize performance, reduce memory usage, and write cleaner, more readable code.

This article will guide you through the myriad of ways to access entries in NumPy arrays; from basic indexing and slicing in one-dimensional arrays to more advanced techniques like boolean and fancy indexing. Whether you’re a beginner looking to grasp the basics or an experienced coder aiming to refine your skills, this guide will give you the knowledge you need to manipulate NumPy arrays effectively.

Accessing Elements in a 1D Array

When you’re working with NumPy, one of the first things you’ll need to master is accessing specific elements within an array. If you’re dealing with a one-dimensional (1D) array, this is pretty straightforward. NumPy arrays (like anything in Python) are zero-indexed, meaning the first element is at index 0, the second element is at index 1, and so on.

Basic Indexing

To access an element in a 1D array, you just use the index of that element inside square brackets. Let’s start with a simple example:

import numpy as np
# Create a 1D NumPy array
arr = np.array([10, 20, 30, 40, 50])
# Access the first element (index 0)
first_element = arr[0]  # Output: 10
# Access the third element (index 2)
third_element = arr[2]  # Output: 30

Easy, right? You can also use negative indexing to access elements from the end of the array. For example, -1 gives you the last element, -2 gives you the second-to-last, and so on.

# Access the last element using negative indexing
last_element = arr[-1]   # Output: 50
# Access the second-to-last element
second_last = arr[-2]    # Output: 40

Slicing

What if you want to access more than one element at a time? Similar to Python lists, that’s where slicing comes in. Slicing allows you to grab a range of elements by specifying a start, stop, and step inside the brackets. The syntax doesn’t look all that different from how you’d do it with lists and it like this:

arr[start:stop:step]

The default step is 1 if you don’t provide one.

Here’s a quick example of slicing:

# Access elements from index 1 to 3 (exclusive)
slice1 = arr[1:4]     # Output: array([20, 30, 40])
# Access the first three elements
slice2 = arr[:3]      # Output: array([10, 20, 30])
# Access every other element
slice3 = arr[::2]     # Output: array([10, 30, 50])

Notice that in arr[1:4], the slicing stops before the index 4, so you get elements from index 1 to 3. As mentioned in older posts, this “exclusive” nature of the stop index is something you’ll see a lot in Python.

Here are a couple more practical examples of indexing and slicing:

Extracting a Subset of Data: Let’s say you have a large array, and you want to grab just a portion of it:

data = np.array([5, 10, 15, 20, 25, 30, 35, 40])
# Get elements from index 2 to 5
subset = data[2:6]  # Output: array([15, 20, 25, 30])

Reversing an Array

In case you want to reverse a NumPy array,  you can do it easily using slicing with a step of -1:

reversed_data = data[::-1]  # Output: array([40, 35, 30, 25, 20, 15, 10, 5])

Understanding how to access elements in a 1D array is a key first step when working with NumPy. Once you get the hang of it, you’ll be able to pull out exactly the data you need with ease, whether it’s single elements or entire slices.

Next up, we’ll dive into working with 2D arrays, where things get just a little more interesting.

Accessing Elements in a 2D Array

When you start working with two-dimensional (2D) arrays, the concepts of indexing and slicing get a bit more interesting. A 2D array can be thought of as an array of arrays — essentially a matrix with rows and columns. Accessing elements in a 2D array is similar to what we’ve seen in 1D arrays, but now you have to specify both the row and the column.

Row and Column Indexing

In a 2D array, you use a [row, column] notation to access elements. The first index corresponds to the row, and the second index corresponds to the column. Let’s start with a simple example:

import numpy as np
# Create a 2D NumPy array (matrix)
arr_2d = np.array([[1, 2, 3],
                   [4, 5, 6],
                   [7, 8, 9]])
# Access element at row 0, column 1
element = arr_2d[0, 1]  # Output: 2
# Access element at row 2, column 2
element = arr_2d[2, 2]  # Output: 9

Pretty straightforward. You just specify the row and column index within the square brackets. As with 1D arrays, you can also use negative indexing to access elements from the end of the rows or columns.

# Access the last element in the last row
element = arr_2d[-1, -1]  # Output: 9
# Access the second-to-last element in the second-to-last row
element = arr_2d[-2, -2]  # Output: 5
Slicing in 2D Arrays

Just like in 1D arrays, you can slice 2D arrays to access ranges of elements. But now you have two dimensions to work with — rows and columns. You can slice rows, columns, or even entire submatrices. The basic syntax is

arr_2d[row_start:row_stop, col_start:col_stop].

Let’s go through a few examples to see how it works:

# Access the first two rows and the last two columns
submatrix = arr_2d[:2, 1:3]  # Output: array([[2, 3], [5, 6]])

In this example, :2 means “give me the first two rows,” and 1:3 means “give me the second and third columns.” The result is a 2×2 submatrix.

You can also slice entire rows or columns:

# Get the second row
row = arr_2d[1, :]  # Output: array([4, 5, 6])
# Get the third column
column = arr_2d[:, 2]  # Output: array([3, 6, 9])

The colon (:) alone means “take all elements along that axis,” so arr_2d[1, :] grabs all the columns from the second row, and arr_2d[:, 2] grabs all the rows from the third column.

Here are a couple more practical examples of working with 2D arrays:

Extracting a Submatrix: If you want to grab a block of elements from a matrix, slicing makes it really easy:

matrix = np.array([[10, 20, 30, 40],
                   [50, 60, 70, 80],
                   [90, 100, 110, 120]])
# Get a submatrix from rows 1 to 2 and columns 1 to 3
submatrix = matrix[1:3, 1:4]  # Output: array([[ 60,  70,  80], [100, 110, 120]])

Accessing Diagonal Elements

To access the diagonal elements of a 2D array, NumPy provides a convenient diagonal() method:

diagonal = matrix.diagonal()  # Output: array([ 10,  60, 110])

Reversing Rows or Columns: You can also reverse the rows or columns of a 2D array just like in a 1D array:

# Reverse the order of the rows
reversed_rows = matrix[::-1, :]  # Output: array([[ 90, 100, 110, 120],
                          #                     [ 50,  60,  70,  80],
                          #                     [ 10,  20,  30,  40]])

 

# Reverse the order of the columns
reversed_columns = matrix[:, ::-1]  # Output: array([[ 40,  30,  20,  10],
                             #                     [ 80,  70,  60,  50],
                             #                     [120, 110, 100,  90]])

Combining Slicing and Indexing

You can also combine slicing and indexing to access specific sections of your 2D arrays. For example, you might want to get specific rows and specific columns at the same time:

# Create a 2D array
arr_2d = np.array([[1, 2, 3],
                   [4, 5, 6],
                   [7, 8, 9]])
# Access rows 0 to 1 (not including row 2) and columns 1 and 2
result = arr_2d[0:2, [1, 2]]  # Output: array([[2, 3], [5, 6]])

The code and the logic here, as you’ve seen, is pretty straightforward. But that’s not all that there is to NumPy array indexing. Some interesting uses of NumPy arrays can be leveraged with more advanced code.

Advanced Indexing Techniques

Once you’re comfortable with basic indexing and slicing, it’s time to explore some of NumPy’s more powerful and flexible features: boolean indexing and fancy indexing. These techniques allow you to access and manipulate specific subsets of an array based on conditions or custom index lists. Let’s dive in.

Boolean Indexing

Boolean indexing is a handy way to access elements of an array based on a condition. Instead of providing specific indices, you create a boolean mask (an array of True or False values) that marks which elements of the array meet a certain condition. The elements corresponding to True are selected, while those corresponding to False are ignored.

Here’s a simple example:

import numpy as np
# Create a 1D NumPy array
arr = np.array([10, 20, 30, 40, 50])
# Create a boolean mask for elements greater than 25
mask = arr > 25  # Output: array([False, False, True, True, True])
# Use the boolean mask to access elements that satisfy the condition
filtered_arr = arr[mask]  # Output: array([30, 40, 50])

In this example, arr > 25 creates a boolean mask where elements greater than 25 are marked as True. When we apply this mask to the original array, only the elements that meet the condition are returned.

You can also combine conditions using logical operators like & (and) and | (or):

# Access elements that are either greater than 20 or less than 15
complex_condition = (arr > 20) | (arr < 15)  # Output: array([ True, False, True, True, True])
filtered_arr = arr[complex_condition]  # Output: array([10, 30, 40, 50])

Fancy Indexing

Fancy indexing allows you to access multiple specific elements from an array by providing a list or array of indices. This is particularly useful when you need to retrieve non-contiguous elements or elements in a specific order. For instance:

# Create a 1D NumPy array
arr = np.array([10, 20, 30, 40, 50])
# Access elements at indices 0, 2, and 4
selected_elements = arr[[0, 2, 4]]  # Output: array([10, 30, 50])

Fancy indexing works just as well in two-dimensional arrays, allowing you to select specific rows, columns, or individual elements from across the array. For example:

# Create a 2D NumPy array
arr_2d = np.array([[1, 2, 3],
                   [4, 5, 6],
                   [7, 8, 9]])
# Access specific elements at (0, 1), (2, 2), and (1, 0)
selected_elements = arr_2d[[0, 2, 1], [1, 2, 0]]  # Output: array([2, 9, 4])

In this example, we’re picking specific elements from the 2D array using two lists: one for row indices and one for column indices. The result is an array of elements (0,1), (2,2), and (1,0).

Modifying Arrays with Boolean Indexing

You can also use boolean indexing to modify elements of an array. This is a really efficient way to apply changes to specific parts of an array based on conditions, without using loops. Here’s an example:

# Create a 1D NumPy array
arr = np.array([10, 20, 30, 40, 50])
# Set all elements greater than 30 to 100
arr[arr > 30] = 100
print(arr)  # Output: array([ 10,  20,  30, 100, 100])

Here, we replaced all elements greater than 30 with the value 100 by using boolean indexing. No loops necessary.

Fancy Indexing for Modifications

Fancy indexing can also be used for modifying specific elements. This is handy when you want to update several elements at once using a list of indices. For instance:

# Create a 1D NumPy array
arr = np.array([10, 20, 30, 40, 50])
# Modify elements at indices 0, 2, and 4
arr[[0, 2, 4]] = [15, 35, 55]
print(arr)  # Output: array([15, 20, 35, 40, 55])

Combining Boolean and Fancy Indexing

Boolean and fancy indexing can be combined for even more powerful selection and modification. Here’s an example where we use boolean indexing to filter elements, then use fancy indexing to modify the filtered result:

# Create a 1D NumPy array
arr = np.array([10, 20, 30, 40, 50])
# Use boolean indexing to filter elements greater than 20
filtered = arr[arr > 20]  # Output: array([30, 40, 50])
# Use fancy indexing to modify specific elements in the filtered result
filtered[[0, 2]] = [35, 55]
print(filtered)  # Output: array([35, 40, 55])

In this example, we first filter the array using a condition (arr > 20), then modify specific elements from the filtered result using fancy indexing.

Modifying Entries in a NumPy Array

Now that we know how to access specific entries in a NumPy array, the next step is learning how to modify them. Whether you want to change individual elements, entire rows or columns, or even sections of an array, NumPy makes it easy. Let’s go over some of the ways you can modify entries in a NumPy array.

Basic Assignment

You can modify a specific element in a NumPy array just by assigning a new value to it using basic indexing. This is very similar to how we accessed elements in previous sections.

import numpy as np
# Create a 1D NumPy array
arr = np.array([10, 20, 30, 40, 50])
# Modify the element at index 2
arr[2] = 35
print(arr)  # Output: array([10, 20, 35, 40, 50])

You can also modify multiple elements at once using slicing.

# Modify elements at index 1 to 3
arr[1:4] = [25, 45, 55]
print(arr)  # Output: array([10, 25, 45, 55, 50])

Broadcasting

One of the most powerful features of NumPy is broadcasting. Broadcasting allows you to apply operations to entire sections of an array at once, without having to explicitly loop over individual elements. This can greatly improve performance and make your code much more concise.

For instance, you can assign the same value to all elements of a row or column, or perform mathematical operations on entire sections of an array at once. Here’s an example:

# Create a 2D array
arr_2d = np.array([[1, 2, 3],
                   [4, 5, 6],
                   [7, 8, 9]])
# Set all elements in the second row to 10
arr_2d[1, :] = 10
print(arr_2d)
# Output:
# array([[ 1,  2,  3],
#        [10, 10, 10],
#        [ 7,  8,  9]])

In this example, the entire second row is updated in one go using broadcasting.

You can also use broadcasting to apply more complex operations across slices of an array. For example, adding or multiplying all elements in a row or column by a scalar:

# Multiply all elements in the first column by 2
arr_2d[:, 0] *= 2
print(arr_2d)
# Output:
# array([[ 2,  2,  3],
#        [20, 10, 10],
#        [14,  8,  9]])

In this case, the first column of arr_2d is multiplied by 2, and the result is applied to all the elements in that column.

Modifying Entries Using Boolean Indexing

As we saw earlier, boolean indexing is a powerful tool to select specific elements based on a condition. It’s also a great way to modify specific elements of an array without using loops. This method allows you to change all the elements that meet a condition in one go.

# Create a NumPy array
arr = np.array([10, 20, 30, 40, 50])
# Set all elements greater than 30 to 100
arr[arr > 30] = 100
print(arr)  # Output: array([ 10,  20,  30, 100, 100])

In this case, we replaced all elements greater than 30 with the value 100. This can be really helpful when you’re working with large datasets and need to apply conditional modifications.

Modifying Entries Using Fancy Indexing

Fancy indexing can also be used to modify multiple specific elements in an array. It’s particularly useful when you need to update elements in non-contiguous positions. For instance:

# Create a NumPy array
arr = np.array([10, 20, 30, 40, 50])
# Modify elements at indices 0, 2, and 4
arr[[0, 2, 4]] = [15, 35, 55]
print(arr)  # Output: array([15, 20, 35, 40, 55])

Here, we’re using fancy indexing to update the elements at indices 0, 2, and 4 simultaneously. This saves you the trouble of having to modify each element individually.

Modifying Entire Rows or Columns

NumPy makes it easy to modify entire rows or columns. By combining slicing and broadcasting, you can quickly change large parts of an array in a single operation.

# Create a 2D NumPy array
arr_2d = np.array([[1, 2, 3],
                   [4, 5, 6],
                   [7, 8, 9]])
# Set all values in the third column to 99
arr_2d[:, 2] = 99
print(arr_2d)
# Output:
# array([[ 1,  2, 99],
#        [ 4,  5, 99],
#        [ 7,  8, 99]])

In this example, all values in the third column are updated to 99 in one go. Similarly, you can modify entire rows or even submatrices.

Hopefully, these uses and examples of NumPy array accessing showed you the versatility and utility o the package for a wide range of scenarios. As we go deeper into data analytics, you’ll start to see the many applications of the lessons shared here. For now, practice the creation and manipulation of arrays to get yourself familiar with the coding mechanics and syntax.

About Glen Dimaandal

Picture of Glen Dimaandal
Glen Dimaandal is a data scientist from the Philippines. He has a post-graduate degree in Data Science and Business Analytics from the prestigious McCombs School of Business in the University of Texas, Austin. He has nearly 20 years of experience in the field as he worked with major brands from the US, UK, Australia and the Asia-Pacific. Glen is also the CEO of SearchWorks.PH, the Philippines’ most respected SEO agency.
Picture of Glen Dimaandal
Glen Dimaandal is a data scientist from the Philippines. He has a post-graduate degree in Data Science and Business Analytics from the prestigious McCombs School of Business in the University of Texas, Austin. He has nearly 20 years of experience in the field as he worked with major brands from the US, UK, Australia and the Asia-Pacific. Glen is also the CEO of SearchWorks.PH, the Philippines’ most respected SEO agency.
ARTICLE & NEWS

Check our latest news

In data science, saving progress is essential. Just like saving your progress in a video game…

In our last lesson, we introduced the concept of Python packages and NumPy in particular. Short…

Now that we have a solid handle on basic Python programming, we can move on to…

Ready to get started?

Reveal the untapped potential of your data. Start your journey towards data-driven decision making with Griffith Data Innovations today.