7.

1 Introduction

NumPy (Numerical Python) Library

First appeared in 2006 and is the preferred Python array implementation.
High-performance, richly functional n-dimensional array type called ndarray .
Written in C and up to 100 times faster than lists.
Critical in big-data processing, AI applications and much more.
According to libraries.io , over 450 Python libraries depend on NumPy.
Many popular data science libraries such as Pandas, SciPy (Scientific Python) and Keras (for deep learning) are built on or depend on NumPy.

Array-Oriented Programming
Functional-style programming with internal iteration makes array-oriented manipulations concise and straightforward, and reduces the possibility of error.

©1992–2020 by Pearson Education, Inc. All Rights Reserved. This content is based on Chapter 5 of the book Intro to Python for Computer Science and Data Science:
Learning to Program with AI, Big Data and the Cloud (https://amzn.to/2VvdnxE).

DISCLAIMER: The authors and publisher of this book have used their best efforts in preparing the book. These efforts include the development, research, and testing of the
theories and programs to determine their effectiveness. The authors and publisher make no warranty of any kind, expressed or implied, with regard to these programs or to
the documentation contained in these books. The authors and publisher shall not be liable in any event for incidental or consequential damages in connection with, or arising
out of, the furnishing, performance, or use of these programs.

/
7.2 Creating array s from Existing Data
Creating an array with the array function
Argument is an array or other iterable
Returns a new array containing the argument’s elements

In [1]: import numpy as np

In [2]: numbers = np.array([2, 3, 5, 7, 11])

In [3]: type(numbers)

Out[3]: numpy.ndarray

In [4]: numbers

Out[4]: array([ 2, 3, 5, 7, 11])

Multidimensional Arguments

In [5]: np.array([[1, 2, 3], [4, 5, 6]])

Out[5]: array([[1, 2, 3],

[4, 5, 6]])

©1992–2020 by Pearson Education, Inc. All Rights Reserved. This content is based on Chapter 5 of the book Intro to Python for Computer Science and Data Science:
Learning to Program with AI, Big Data and the Cloud (https://amzn.to/2VvdnxE).

DISCLAIMER: The authors and publisher of this book have used their best efforts in preparing the book. These efforts include the development, research, and testing of the
theories and programs to determine their effectiveness. The authors and publisher make no warranty of any kind, expressed or implied, with regard to these programs or to
the documentation contained in these books. The authors and publisher shall not be liable in any event for incidental or consequential damages in connection with, or arising
out of, the furnishing, performance, or use of these programs.

/
7.3 array Attributes
attributes enable you to discover information about its structure and contents

In [1]: import numpy as np

In [2]: integers = np.array([[1, 2, 3], [4, 5, 6]])

In [3]: integers

Out[3]: array([[1, 2, 3],

[4, 5, 6]])

In [4]: floats = np.array([0.0, 0.1, 0.2, 0.3, 0.4])

In [5]: floats

Out[5]: array([0. , 0.1, 0.2, 0.3, 0.4])

NumPy does not display trailing 0s

Determining an array ’s Element Type

In [6]: integers.dtype

Out[6]: dtype('int64')

In [7]: floats.dtype

Out[7]: dtype('float64')

For performance reasons, NumPy is written in the C programming language and uses C’s data types
Other NumPy types (https://docs.scipy.org/doc/numpy/user/basics.types.html)

Determining an array ’s Dimensions

ndim contains an array ’s number of dimensions
shape contains a tuple specifying an array ’s dimensions

In [8]: integers.ndim

Out[8]: 2

In [9]: floats.ndim

Out[9]: 1

In [10]: integers.shape

Out[10]: (2, 3)

In [11]: floats.shape

Out[11]: (5,)

Determining an array ’s Number of Elements and Element Size

view an array ’s total number of elements with size
view number of bytes required to store each element with itemsize

/
 In [12]: integers.size

Out[12]: 6

In [13]: integers.itemsize

Out[13]: 8

In [14]: floats.size

Out[14]: 5

In [15]: floats.itemsize

Out[15]: 8

Iterating through a Multidimensional array ’s Elements

In [16]: for row in integers:

for column in row:
print(column, end=' ')
print()

1 2 3
4 5 6

Iterate through a multidimensional array as if it were one-dimensional by using flat

In [17]: for i in integers.flat:

print(i, end=' ')

1 2 3 4 5 6

©1992–2020 by Pearson Education, Inc. All Rights Reserved. This content is based on Chapter 5 of the book Intro to Python for Computer Science and Data Science:
Learning to Program with AI, Big Data and the Cloud (https://amzn.to/2VvdnxE).

DISCLAIMER: The authors and publisher of this book have used their best efforts in preparing the book. These efforts include the development, research, and testing of the
theories and programs to determine their effectiveness. The authors and publisher make no warranty of any kind, expressed or implied, with regard to these programs or to
the documentation contained in these books. The authors and publisher shall not be liable in any event for incidental or consequential damages in connection with, or arising
out of, the furnishing, performance, or use of these programs.

/
7.4 Filling array s with Specific Values
Functions zeros , ones and full create array s containing 0 s, 1 s or a specified value, respectively

In [1]: import numpy as np

In [2]: np.zeros(5)

Out[2]: array([0., 0., 0., 0., 0.])

For a tuple of integers, these functions return a multidimensional array with the specified dimensions

In [3]: np.ones((2, 4), dtype=int)

Out[3]: array([[1, 1, 1, 1],

[1, 1, 1, 1]])

In [4]: np.full((3, 5), 13)

Out[4]: array([[13, 13, 13, 13, 13],

[13, 13, 13, 13, 13],
[13, 13, 13, 13, 13]])

©1992–2020 by Pearson Education, Inc. All Rights Reserved. This content is based on Chapter 5 of the book Intro to Python for Computer Science and Data Science:
Learning to Program with AI, Big Data and the Cloud (https://amzn.to/2VvdnxE).

DISCLAIMER: The authors and publisher of this book have used their best efforts in preparing the book. These efforts include the development, research, and testing of the
theories and programs to determine their effectiveness. The authors and publisher make no warranty of any kind, expressed or implied, with regard to these programs or to
the documentation contained in these books. The authors and publisher shall not be liable in any event for incidental or consequential damages in connection with, or arising
out of, the furnishing, performance, or use of these programs.

/
7.10 Indexing and Slicing
One-dimensional array s can be indexed and sliced like lists.

Indexing with Two-Dimensional array s

To select an element in a two-dimensional array , specify a tuple containing the element’s row and column indices in square brackets

In [1]: import numpy as np

In [2]: grades = np.array([[87, 96, 70], [100, 87, 90],

[94, 77, 90], [100, 81, 82]])

In [3]: grades

Out[3]: array([[ 87, 96, 70],

[100, 87, 90],
[ 94, 77, 90],
[100, 81, 82]])

In [4]: grades[0, 1] # row 0, column 1

Out[4]: 96

Selecting a Subset of a Two-Dimensional array ’s Rows

To select a single row, specify only one index in square brackets

In [5]: grades[1]

Out[5]: array([100, 87, 90])

Select multiple sequential rows with slice notation

In [6]: grades[0:2]

Out[6]: array([[ 87, 96, 70],

[100, 87, 90]])

Select multiple non-sequential rows with a list of row indices

In [7]: grades[[1, 3]]

Out[7]: array([[100, 87, 90],

[100, 81, 82]])

Selecting a Subset of a Two-Dimensional array ’s Columns

The column index also can be a specific index, a slice or a list

In [8]: grades[:, 0]

Out[8]: array([ 87, 100, 94, 100])

In [9]: grades[:, 1:3]

Out[9]: array([[96, 70],

[87, 90],
[77, 90],
[81, 82]])

/
 In [10]: grades[:, [0, 2]]

Out[10]: array([[ 87, 70],

[100, 90],
[ 94, 90],
[100, 82]])

©1992–2020 by Pearson Education, Inc. All Rights Reserved. This content is based on Chapter 5 of the book Intro to Python for Computer Science and Data Science:
Learning to Program with AI, Big Data and the Cloud (https://amzn.to/2VvdnxE).

DISCLAIMER: The authors and publisher of this book have used their best efforts in preparing the book. These efforts include the development, research, and testing of the
theories and programs to determine their effectiveness. The authors and publisher make no warranty of any kind, expressed or implied, with regard to these programs or to
the documentation contained in these books. The authors and publisher shall not be liable in any event for incidental or consequential damages in connection with, or arising
out of, the furnishing, performance, or use of these programs.