NumPy Recipes

By Martin McBride

NumPy Recipes takes practical approach to the basics of NumPy.

 

This book is primarily aimed at developers who have at least a small amount of Python experience, who wish to use the NumPy library for data analysis, machine learning, image or sound processing, or any other mathematical or scientific application. It only requires a basic understanding of Python programming.

 

Detailed examples show how to create arrays to optimise storage different types of information, and how to use universal functions, vectorisation, broadcasting and slicing to process data efficiently. Also contains an introduction to file i/o and data visualisation with Matplotlib.

• Language: English
• Publisher: Martin McBride
• Release date: Sep 23, 2020
• ISBN: 9781393646075

Book preview

NumPy Recipes

NumPy Recipes

A practical introduction to NumPy

Martin McBride

© 2020 Martin McBride

Table of Contents

Foreword

Who is this book for?

About the author

Keep in touch

1 Introduction to NumPy

1.1 Installing NumPy

1.2 What is NumPy?

1.3 NumPy vs Python lists

1.4 Advantages of NumPy

1.5 NumPy universal functions

1.6 Compatibility with other libraries

2 Anatomy of a NumPy array

2.1 NumPy arrays compared to lists

2.2 Printing the characteristics of an array

2.2.1 Rank

2.2.2 Shape

2.2.3 Size

2.2.4 Data type

2.2.5 Item size

2.2.6 Data location

2.3 Array rank examples

2.4 Data types

2.4.1 Integers

2.4.2 Unsigned integers

2.4.3 Floating point values

2.4.4 Complex number formats

2.4.5 Boolean

2.4.6 List of main data types

3 Creating arrays

3.1 Creating an array of zeroes

3.2 Creating other fixed content arrays

3.3 Choosing the data type

3.4 Creating multi-dimensional arrays

3.5 Creating like arrays

3.6 Creating an array from a Python list

3.7 Controlling the type with the array function

3.8 array function anti-patterns

3.9 Creating a value series with arange

3.10 Rounding error problem with arange

3.11 Create a sequence of a specific length with linspace

3.12 Making linspace more like arange using the endpoint parameter

3.13 Obtaining the linspace step size

3.14 Other sequence generators

3.15 Creating an identity matrix

3.16 Creating an eye matrix

3.17 Using vectorisation

4 Vectorisation

4.1 Performing simple maths on an array

4.2 Vectorisation with other data types

4.3 Vectorisation with multi-dimensional arrays

4.4 Expressions using two arrays

4.5 Expressions using two multi-dimensional arrays

4.6 More complex expressions

4.7 Using conditional operators

4.8 Combining conditional operators

5 Universal functions

5.1 Example universal function - sqrt

5.2 Example universal function of two arguments - power

5.3 Summary of ufuncs

5.3.1 Maths operations

5.3.2 Trigonometric functions

5.3.3 Bit manipulation

5.3.4 Comparison functions

5.3.5 Logical functions

5.3.6 Min and max

5.3.7 Float functions

5.4 ufunc methods

5.4.1 Reduce

5.4.2 Accumulation

5.5 Optional keyword arguments for ufuncs

5.5.1 out

5.5.2 where

6 Indexing, slicing and broadcasting

6.1 Indexing an array

6.1.1 Indexing in 1 dimension

6.1.2 Indexing in 2 dimensions

6.1.3 Picking a row or column in 2-dimensions

6.1.4 Indexing in 3 dimensions

6.1.5 Picking a row or column in a 3D array

6.1.6 Picking a matrix in a 3D array

6.2 Slicing an array

6.2.1 Slicing lists - a recap

6.2.2 Slicing 1D NumPy arrays

6.2.3 Slicing a 2D array

6.2.4 Slicing a 3D array

6.2.5 Full slices

6.3 Slices vs indexing

6.4 Views

6.5 Broadcasting

6.5.1 Broadcasting from 1 to 2 dimensions

6.5.2 Broadcasting 1 to 3 dimensions

6.5.3 Broadcasting 2 to 3 dimensions

6.6 Broadcasting rules

6.7 Broadcasting a column vector

6.8 Broadcasting a row vector and a column vector

6.9 Broadcasting scalars

6.10 Efficient broadcasting

6.11 Fancy indexing

7 Array manipulation functions

7.1 Copying an array

7.2 Changing the type of an array

7.3 Changing the shape of an array

7.4 Splitting arrays

7.4.1 Splitting along different axes

7.4.2 Unequal splits

7.4.3 Alternative functions

7.5 Stacking arrays

7.5.1 Stacking 2-dimensional arrays

8 File input and output

8.1 CSV format

8.2 Writing CSV data

8.2.1 Adding a header or footer

8.2.2 Changing the line separator

8.2.3 Compressing the output file

8.3 Reading CSV data

8.3.1 Skipping header or footer

8.3.2 Reading compressing file

9 Using Matplotlib with NumPy

9.1 Installing Matplotlib

9.2 Plotting a histogram

9.3 Plotting functions

9.4 Plotting functions with NumPy

9.5 Creating a heatmap

10 Reference

10.1 Data types

10.1.1 Unsigned integer sizes and ranges

10.1.2 Signed integer sizes and ranges

10.1.3 Integer wrap-around

10.1.4 Float characteristics

10.1.5 Complex characteristics

Foreword

This book aims to provide an introduction to the use and application of the Python NumPy library. NumPy is a large library, and this book introduces many of its most important features in the form of recipes for performing certain functions. This provides a practical grounding that can be used as a basis for understanding the extensive documentation available from the NumPy website.

Who is this book for?

This book is primarily aimed at developers who have at least a small amount of Python experience, who wish to use the NumPy library for data analysis, machine learning, image or sound processing, or any other mathematical or scientific application. It only requires a basic understanding of Python programming.

About the author

Martin McBride is a software developer, specialising in computer graphics, sound, and mathematical programming. He has been writing code since the 1980s in a wide variety of languages from assembler through to C++, Java and Python. He writes for PythonInformer.com and is the author of Functional Programming in Python available on leanpub.com. He is interested in generative art and works on the generativepy open source project.

Keep in touch

If you have any comments or questions you can get in touch by any of the following methods:

Joining the Python Informer forum at http://pythoninformer.boards.net/. Follow the NumPy board in the Books section.

Signing up for the Python Informer newsletter at pythoninformer.com

Following @pythoninformer on twitter.

Contacting me directly by email ([e-mail]).

1 Introduction to NumPy

This is a brief overview of what NumPy is, and why makes it so useful. But first, we will