Contiguous arrays and stride tricks in NumPy - Mini Project: Build & Apply

Contiguous Arrays and Stride Tricks with NumPy

📖 Scenario: You work with sensor data collected every second. The data is stored in a NumPy array. You want to analyze small windows of 3 seconds each without copying data to save memory and time.

🎯 Goal: Learn how to create a sliding window view of a NumPy array using stride tricks to get contiguous subarrays efficiently.

📋 What You'll Learn

Create a NumPy array with exact values

Define a window size variable

Use as_strided from numpy.lib.stride_tricks to create sliding windows

Print the resulting array of windows

💡 Why This Matters

🌍 Real World

Sliding window views are useful in signal processing, time series analysis, and machine learning to analyze data chunks efficiently without extra memory.

💼 Career

Data scientists and engineers often need to process large datasets efficiently. Understanding stride tricks helps optimize performance and memory usage.

1

Create the sensor data array

Create a NumPy array called sensor_data with these exact values: [10, 20, 30, 40, 50, 60].

NumPy

# Create the sensor_data array with the given values
# Your code here

Need a hint?

Use np.array to create the array with the exact values.

2

Set the window size

Create a variable called window_size and set it to 3.

NumPy

import numpy as np
sensor_data = np.array([10, 20, 30, 40, 50, 60])
# Set the window_size variable to 3
# Your code here

Need a hint?

Just assign the number 3 to window_size.

3

Create sliding windows using stride tricks

Import as_strided from numpy.lib.stride_tricks and use it to create a variable called windows that contains sliding windows of size window_size from sensor_data. Use the strides of sensor_data and set the shape to get all possible windows.

NumPy

import numpy as np
sensor_data = np.array([10, 20, 30, 40, 50, 60])
window_size = 3
# Create sliding windows using as_strided
# Your code here

Need a hint?

Use as_strided with shape (sensor_data.size - window_size + 1, window_size) and strides equal to sensor_data.strides[0] for both dimensions.

4

Print the sliding windows

Print the variable windows to see the sliding windows of the sensor data.

NumPy

import numpy as np
sensor_data = np.array([10, 20, 30, 40, 50, 60])
window_size = 3
from numpy.lib.stride_tricks import as_strided
windows = as_strided(sensor_data, shape=(sensor_data.size - window_size + 1, window_size), strides=(sensor_data.strides[0], sensor_data.strides[0]))
# Print the windows variable
# Your code here

Need a hint?

Use print(windows) to display the sliding windows.