Business Analytics

NumPy Basics

Phan Xuân Hiếu • VNU-UET • [email protected]

Lecture objectives

§ Introduce the NumPy library and explain why NumPy is important to data analytics.
§ Introduce and describe the NumPy multidimensional arrays (ndarray) with
– Different ways to create arrays,
– The standard data types,
– Different array indexing and slicing methods, and
– Various operations on arrays including arithmetic, linear algebra, aggregate, etc,
– Ways to reshape, concatenate, and split arrays.

§ Introduce various universal functions working on NumPy arrays.

§ Introduce how to save and load NumPy arrays to and from files.

20 March 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 2

Contents

1. What Is NumPy?
2. How Python and NumPy Store Data Objects
3. The NumPy ndarray: A Multidimensional Array Object
4. Pseudorandom Number Generation
5. Universal Functions: Fast Element-Wise Array Functions
6. Array-Oriented Programming with Arrays
7. File Input and Output with Arrays

20 March 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 3

What Is NumPy?
– What Is NumPy?
– Why Is NumPy?

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 4
What is NumPy?
§ Created in 2005 by Travis Oliphant.
– It is an open source project and you can use it freely.

§ NumPy (short for Numerical Python):

– One of the most important foundational package for numerical computing in Python.

§ Many important libraries built on top of or with NumPy:

– SciPy, pandas, matplotlib, seaborn, scikit-learn, TensorFlow, PyTorch ...

§ The main element of NumPy is (multidimensional) arrays.

– Fixed-typed arrays (homogeneous).
– Efficient memory useage and fast operations on arrays.

§ Also provides functions in linear algebra, fourier transform, random ...

20 March 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 5

Technically, ... things you will find in NumPy

§ ndarray, an efficient multidimensional array providing

– fast array-oriented arithmetic operations and flexible broadcasting capabilities

§ Mathematical functions for fast operations on entire arrays of data

– without having to write loops

§ Tools for reading and writing array data to disk

– and working with memory-mapped files.

§ Linear algebra, random number generation, and Fourier transform capabilities

§ C APIs for connecting NumPy with libraries written in C, C++, FORTRAN

20 March 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 6

Why is NumPy?
§ Understanding NumPy:
– working with arrays and other libraries like pandas, scikit-learn more effectively
– a lot of NumPy operations can be applied in data analysis applications.

§ We will focus on:

– Fast array-based operations for data munging, cleaning, transformation ...
– Common array algorithms like sorting, unique, and set operations ...
– Efficient descriptive statistics and aggregation / summarizing data
– Data alignment and relational data manipulations for merging, joining data ...
– Expressing conditional logic as array expressions instead of loops
– Group-wise data manipulations ...

20 March 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 7

How Python and NumPy Store Data Objects
– A Python Integer is More Than Just an Integer
– A Python List Is More Than Just a List
– Fixed-Type Arrays in Python
– NumPy Arrays versus Python Lists

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 8
Why need to understand data types in Python and NumPy?

§ Data science requires understanding how data is stored and manipulated.

§ Constrats:
– How arrays of data are handled in the Python language itself, and
– How NumPy improves on this.

§ Python is a dynamically typed language

– (vs. statically typed languages like C or Java)
– Python’s type flexibility → requires more memory to store variables ...

§ And how NumPy organize and store data to be fast and efficient?

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 9
A Python integer is more than just an integer

A single integer in Python 3.10 (C structure)

§ ob_refcnt: a reference count helping Python silently
handle memory allocation and deallocation
§ ob_type: the pointer to the Python type object that
defines behavior (should be <class ‘int’>)
§ ob_size: the number of digits stored in ob_digit
§ ob_digit: the actual integer value, can grow as needed.
Python integer is not limited by CPU bit size (unlike C
int). It is arbitrary precision.

20 March 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 10

C integer versus Python integer

§ Some overhead in storing integer in Python

§ PyObject_HEAD: the extra data (e.g.,
reference count, object type ...)
§ A Python integer is a pointer to a position in
memory containing all the Python object
information as explained.
§ Dynamic typing in Python comes at a cost!

20 March 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 11

A Python list is more than just a list

§ A Python list can contain objects of different types (i.e., heterogeneous)

§ Each item in the list must contain its own type, reference count ...
§ Each item is a complete Python object.
§ If all elements of the list are of the same type → much redundancy!
§ Much more efficient to store the data in a fixed-type array.

20 March 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 12

Fixed-type arrays in Python

§ The built-in array module.

§ Array of a uniform type.
§ ‘i’ is a type code → integer.
§ However:
• NumPy arrays are more
efficient (both for storing
and efficient operations)

20 March 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 13

NumPy arrays vs. Python lists
§ Python list:
• Heterogeneous (objects of different
types) → flexibility but pay a cost!
• Pointer → a block of pointers →
pointing to full Python objects (like
Python integer as described)
§ NumPy array:
• Fixed-type, i.e., homogeneous (all
elements are of the same type)
• Not flexible as Python list but
efficient for storing and
manipulating data.

20 March 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 14

The NumPy ndarray: A Multidimensional Array Object
– Creating NumPy Arrays – Arithmetic Operations
– NumPy Standard Data Types – Linear Algebra Operations
– Array Indexing and Accessing – Aggregate Functions
– Slicing Subarrays – Shape Manipulation Operations
– Fancy Indexing – Array Concatenation and Splitting

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 15
The NumPy ndarray
§ One of the key features of NumPy:
– its n-dimensional array object, or ndarray.
– ndarray is a fast, flexible container for large datasets in Python.
– ndarrays allow math operations on whole blocks of data ...

§ Import of NumPy and creating a simple 2-dimensional array (i.e., a matrix)

20 March 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 16

narray – multidimensional container for homogeneous data
§ Homogeneous: all elements must be the same type.
§ Every array has a ndim: an integer indicating the number of dimensions
§ Every array has a shape: a tuple indicating the size of each dimension.
§ A dtype: an object describing the data type of the array.

20 March 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 17

The NumPy ndarray: A Multidimensional Array Object
– Creating NumPy Arrays – Arithmetic Operations
– NumPy Standard Data Types – Linear Algebra Operations
– Array Indexing and Accessing – Aggregate Functions
– Slicing Subarrays – Shape Manipulation Operations
– Fancy Indexing – Array Concatenation and Splitting

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 18
Creating NumPy arrays
§ Creating 0-dimensional arrays
§ Creating arrays from Python sequences (lists, tuples ...)
§ Creating arrays with default values
§ Creating arrays with functions arange() or linspace()
§ Creating arrays with random values (uniform / normal distributions)
§ Creating an uninitialized array using np.empty()
§ Creating multidimensional arrays using reshape()
§ Creating 1-dimensional arrays using reshape(-1)

20 March 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 19

Creating 0-dimensional arrays (vs. 1-dimensional arrays)

20 March 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 20

Creating arrays from Python sequences (lists, tuples ...)

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 21
Check and specify data type when creating arrays

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 22
Creating arrays with default values

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 23
Creating arrays with arange() and linspace()

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 24
Some important NumPy array creation functions

20 March 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 25

Creating arrays of random values in [0, 1)

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 26
Creating arrays of random integers in a given range

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 27
Creating arrays of values from normal distributions

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 28
Creating arrays of values from normal distributions (2)

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 29
Creating an uninitialized array using np.empty()

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 30
Creating multidimensional arrays using reshape()

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 31
Flattening 1-dimensional arrays using reshape(-1)

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 32
The NumPy ndarray: A Multidimensional Array Object
– Creating NumPy Arrays – Arithmetic Operations
– NumPy Standard Data Types – Linear Algebra Operations
– Array Indexing and Accessing – Aggregate Functions
– Slicing Subarrays – Shape Manipulation Operations
– Fancy Indexing – Array Concatenation and Splitting

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 33
NumPy standard data types
§ NumPy arrays contain values of a single type.
§ NumPy is built in C
– The types will be familiar to users of C, Fortran, and other related languages.
§ Can specify the data type when creating a NumPy array

20 March 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 34

20 March 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 35
Type conversion or casting for arrays

20 March 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 36

Type conversion or casting for arrays (2)

20 March 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 37

The NumPy ndarray: A Multidimensional Array Object
– Creating NumPy Arrays – Arithmetic Operations
– NumPy Standard Data Types – Linear Algebra Operations
– Array Indexing and Accessing – Aggregate Functions
– Slicing Subarrays – Shape Manipulation Operations
– Fancy Indexing – Array Concatenation and Splitting

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 38
Array indexing and accessing elements

§ Accessing single elements

§ Indexing errors
§ Updating elements

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 39
Accessing single elements in 1-d arrays (i.e., vectors)

0 1 2 3 4

arr1d = 2 8 3 3 0

-5 -4 -3 -2 -1

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 40
Accessing single elements in 1-d arrays with negative indices

0 1 2 3 4

arr1d = 2 8 3 3 0

-5 -4 -3 -2 -1

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 41
Accessing single elements in 2-d arrays (i.e., matrices)

arr2d =

Axis 1
0 1 2 3

0 18 12 3 19 -3
Axis 0

1 12 0 5 10 -2

2 14 1 8 5 -1

-4 -3 -2 -1

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 42
Accessing single elements in 2-d arrays with negative indices

arr2d =

Axis 1
0 1 2 3

0 18 12 3 19 -3
Axis 0

1 12 0 5 10 -2

2 14 1 8 5 -1

-4 -3 -2 -1

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 43
Accessing single elements in 3-d arrays

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 44
Indexing errors in 1-d arrays

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 45
Indexing errors in 2-d arrays

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 46
Updating elements in arrays

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 47
The NumPy ndarray: A Multidimensional Array Object
– Creating NumPy Arrays – Arithmetic Operations
– NumPy Standard Data Types – Linear Algebra Operations
– Array Indexing and Accessing – Aggregate Functions
– Slicing Subarrays – Shape Manipulation Operations
– Fancy Indexing – Array Concatenation and Splitting

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 48
Slicing subarrays from one-dimensional arrays

§ The slicing syntax:

array[start:stop:step]
– start is the the first index (included)
– stop is the last index (excluded)
– |step| > 1 if want to skip some elements at each step
§ The default values:
– start = 0
– stop = <size of dimension>
– step = 1

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 49
Slicing from one-dimensional arrays

0 1 2 3 4

arr1d = 8 6 7 5 9

-5 -4 -3 -2 -1

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 50
Slicing from one-dimensional arrays with negative indices

0 1 2 3 4

arr1d = 8 6 7 5 9

-5 -4 -3 -2 -1

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 51
Slicing subarrays from multidimensional arrays

§ The slicing syntax (k ≥ 2):

array[start1:stop1:step1, ..., startk:stopk:stepk]
– starti is the the first index (included)
– stopi is the last index (excluded)
– |stepi| > 1 if want to skip some elements at each step
§ Default values:
– starti = 0
– stopi = <size of dimension i>
– stepi = 1

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 52
Slicing subarrays from 2-d arrays

arr2d =

Axis 1
0 1 2 3

0 18 12 3 19 -3
Axis 0

1 12 0 5 10 -2

2 14 1 8 5 -1

-4 -3 -2 -1

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 53
Different ways of slicing and their shapes

arr2d = Slicing expression Shape

Axis 1
arr2d[1] (4,)
0 1 2 3
arr2d[1, :] (4,)
0 18 12 3 19 -3
Axis 0

arr2d[-2, :] (4,)
1 12 0 5 10 -2
arr2d[1:2, :] (1, 4)
2 14 1 8 5 -1
arr2d[-2:-1, :] (1, 4)
-4 -3 -2 -1

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 54
Different ways of slicing and their shapes (2)

arr2d = Slicing expression Shape

Axis 1
arr2d[2, 1:3] (2,)
0 1 2 3
arr2d[2, 1:-1] (2,)
0 18 12 3 19 -3
Axis 0

arr2d[-1, -3:-1] (2,)

1 12 0 5 10 -2
arr2d[2:, 1:3] (1, 2)
2 14 1 8 5 -1
arr2d[-1:, 1:-1] (1, 2)
-4 -3 -2 -1

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 55
Different ways of slicing and their shapes (3)

arr2d =

Axis 1
0 1 2 3 Slicing expression Shape

0 18 12 3 19 -3 arr2d[:, :2] (3, 2)

Axis 0

1 12 0 5 10 -2 arr2d[:3, :2] (3, 2)

2 14 1 8 5 -1 arr2d[:, :-2] (3, 2)

-4 -3 -2 -1

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 56
Different ways of slicing and their shapes (4)

arr2d =
Slicing expression Shape
Axis 1
0 1 2 3 arr2d[:2, 2:] (2, 2)

0 18 12 3 19 -3 arr2d[:2, -2:] (2, 2)

Axis 0

1 12 0 5 10 -2 arr2d[:-1, -2:] (2, 2)

2 14 1 8 5 -1 arr2d[0:2, 2:4] (2, 2)

-4 -3 -2 -1

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 57
Slicing subarrays from 3-d arrays

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 58
Slicing subarrays from 3-d arrays (2)

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 59
Subarrays as no-copy views

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 60
Subarrays as no-copy views (2)

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 61
Creating copies of subarrays

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 62
The NumPy ndarray: A Multidimensional Array Object
– Creating NumPy Arrays – Arithmetic Operations
– NumPy Standard Data Types – Linear Algebra Operations
– Array Indexing and Accessing – Aggregate Functions
– Slicing Subarrays – Shape Manipulation Operations
– Fancy Indexing – Array Concatenation and Splitting

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 63
Indexing rows using integer arrays

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 64
Indexing columns using integer arrays

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 65
Indexing particular elements, e.g., (1, 0), (5, 3), (7, 1), (2, 2)

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 66
Selecting a subset of the matrix’s rows and columns

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 67
Modifying the original matrix with fancy indexing

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 68
The NumPy ndarray: A Multidimensional Array Object
– Creating NumPy Arrays – Arithmetic Operations
– NumPy Standard Data Types – Linear Algebra Operations
– Array Indexing and Accessing – Aggregate Functions
– Slicing Subarrays – Shape Manipulation Operations
– Fancy Indexing – Array Concatenation and Splitting

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 69
Arithmetic operations (element-wise)
§ Addition (+)
§ Subtraction (-)
§ Multiplication (*)
§ Division (/)
§ Floor division (//)
§ Modulus (%)
§ Exponentiation (**)

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 70
Arithmetic operations on two vectors (element-wise)

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 71
Arithmetic operations on two matrices (element-wise)

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 72
The NumPy ndarray: A Multidimensional Array Object
– Creating NumPy Arrays – Arithmetic Operations
– NumPy Standard Data Types – Linear Algebra Operations
– Array Indexing and Accessing – Aggregate Functions
– Slicing Subarrays – Shape Manipulation Operations
– Fancy Indexing – Array Concatenation and Splitting

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 73
Linear algebra operations
§ Matrix multiplication: A @ B or np.dot(A, B)
§ Transpose of a matrix: A.T or np.transpose(A)
§ Inverse of a matrix: np.linalg.inv(A)
§ Determinant of a matrix: np.linalg.det(A)
§ Solve linear equations: np.linalg.solve(A, b)
§ Eigenvalues and eigenvectors: np.linalg.eig(A)

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 74
Examples of linear algebra operations

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 75
Examples of linear algebra operations (2)

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 76
Examples of linear algebra operations (3)

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 77
Examples of linear algebra operations (4)

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 78
Commonly used numpy.linalg functions

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 79
The NumPy ndarray: A Multidimensional Array Object
– Creating NumPy Arrays – Arithmetic Operations
– NumPy Standard Data Types – Linear Algebra Operations
– Array Indexing and Accessing – Aggregate Functions
– Slicing Subarrays – Shape Manipulation Operations
– Fancy Indexing – Array Concatenation and Splitting

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 80
Aggregate functions (reduction operations)
§ Sum of elements: np.sum(A) § Minimum value: np.min(A)
§ Mean (average): np.mean(A) § Maximum value: np.max(A)
§ Median: np.median(A) § Product of all elements: np.prod(A)
§ Standard deviation: np.std(A) § Cumulative sum: np.cumsum(A)
§ Variance: np.var(A) § Cumulative product: np.cumprod(A)

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 81
Examples of aggregate functions (reduction operations)

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 82
Examples of aggregate functions (reduction operations) (2)

20 March 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 83

Examples of aggregate functions (reduction operations) (3)

20 March 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 84

The NumPy ndarray: A Multidimensional Array Object
– Creating NumPy Arrays – Arithmetic Operations
– NumPy Standard Data Types – Linear Algebra Operations
– Array Indexing and Accessing – Aggregate Functions
– Slicing Subarrays – Shape Manipulation Operations
– Fancy Indexing – Array Concatenation and Splitting

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 85
Shape manipulation operations

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 86
Reshaping 1D arrays to 2D, 3D arrays ...

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 87
Converting 2D arrays to 1D arrays

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 88
Converting 3D arrays to 1D arrays

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 89
Reshape with -1 (automatically calculate)

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 90
Reshape with -1 (automatically calculate) (2)

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 91
Adding a new axis (expanding dimensions)

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 92
Removing single-dimension axes (squeeze)

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 93
The NumPy ndarray: A Multidimensional Array Object
– Creating NumPy Arrays – Arithmetic Operations
– NumPy Standard Data Types – Linear Algebra Operations
– Array Indexing and Accessing – Aggregate Functions
– Slicing Subarrays – Shape Manipulation Operations
– Fancy Indexing – Array Concatenation and Splitting

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 94
Concatenation of arrays
§ Concatenation: np.concatenate()
§ Vertical stacking: np.vstack()
§ Horizontal stacking: np.hstack()
§ Column-wise stacking: np.column_stack()
§ Row-wise stacking: np.row_stack()

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 95
Array concatenation

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 96
Vertical and horizontal stacking

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 97
Column-wise and row-wise stacking

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 98
Splitting of arrays

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 99
Vertical and horizontal splitting

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 100
Pseudorandom Number Generation
– Generation from Uniform Distribution
– Generation from Normal Distributions
– Generation from Binomial Distributions
– Generating Random Permutations of Elements
– Generation from Other Distributions

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 101
Random generation from uniform distribution

§ Generating random integers from uniform distribution

§ Generating random real numbers in [0, 1) from uniform distribution
§ Generating random numbers from an array
– With uniform distribution over the array’s elements or there are weights
– With replacement (replace=True), and
– Without replacement (replace=False)

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 102
Generating random integers from uniform distribution

§ Generating random integers

from uniform distribution in
the range [low, high), e.g.,
• [0, 100)
§ size=shape, e.g.,
• size=10
• size=(3, 5)
• size=(2, 2, 4)

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 103
Generating random numbers in [0, 1) from uniform dist.

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 104
Generating random numbers from an array

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 105
Generating random numbers from an array (2)

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 106
Random generation from normal distributions

§ The normal distribution is one of the most important distributions.

– Also called the Gaussian distribution (after Carl Friedrich Gauss)
– It fits the probability distribution of many real-world events, e.g., IQ scores, heights ...

§ Three ways to generate numbers from the normal distribution:

– numpy.random.randn(d0, d1, ...)
• Standard normal distribution with shape information
– numpy.random.standard_normal(size=(d0, d1, ...))
• Standard normal distribution with shape information (in a tuple)
– numpy.random.normal(loc=0.0, scale=1.0, size=(d0, d1, ....))
• Normal distribution with mean, standard deviation, and array shape (default = None)

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 107
Random generation from the standard normal distribution

20 March 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 108
Generation from normal distribution (mean=3.0, std=2.0)

20 March 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 109
Plotting the sample histogram from the normal distribution

20 March 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 110
Random generation from the binomial distribution

§ The binomial distribution models the number of successes in a fixed

number of independent trials, where each trial has only two possible
outcomes: success or failure.
– The binomial distribution is a discrete distribution.
– It describes the outcome of binary scenarios, e.g., toss of a coin ...

§ It has two parameters:

– 𝑛 is the total number of independent trials.
– 𝑝 is the probability of success in each trial.

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 111
Relationship between binomial and normal distributions

§ The binomial distribution approaches a normal distribution when:

– 𝑛 (the number of trials) is large.
– 𝑝 (the probability of success) is not too close to 0 or 1.
– Approximation is valid when 𝑛𝑝 ≥ 5 and 𝑛(1 − 𝑝) ≥ 5.

§ The normal distribution approximation to the binomial distribution:

– Mean: 𝜇 = 𝑛𝑝
– Variance: 𝜎 ! = 𝑛𝑝(1 − 𝑝)

20 March 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 112
Example of generation from the binomial distribution

20 March 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 113
Plotting binomial sample distribution (n=10, p=0.2)

20 March 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 114
When the binomial approaching normal (n=100, p=0.5)

20 March 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 115
Generating random permutations of elements

§ Shuffing arrays (in-place shuffle)

§ Generating permutations of the elements of an array (return a copy)

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 116
Shuffing the elements of an array

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 117
Generating permutations of the elements of an array

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 118
Universal Functions: Fast Element-Wise Array Functions
– What Are Universal Functions?
– Create Your Own Ufunc
– Other NumPy Ufuncs

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 119
What are universal functions?

§ A universal function, or ufunc, is a function that

– perform element-wise operations on data in ndarrays.

§ Ufuncs are used to implement vectorization in NumPy

– which is way faster than iterating over elements.

§ Ufuncs also take additional arguments, like:

– where: boolean array or condition defining where the operations should take place.
– dtype: defining the return type of elements.
– out: output array where the return value should be copied.

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 120
How do ufuncs work?

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 121
Creating your own ufuncs

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 122
Ufuncs for simple arithmetic operations
§ Addition: np.add()
§ Subtraction: np.subtract()
§ Multiplication: np.multiply()
§ Division: np.divide()
§ Power: np.power()
§ Remainder: np.mod() or np.remainder()
§ Quotient and mod: np.divmod()
§ Absolute values: np.absolute()

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 123
Examples of arithmetic ufunctions

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 124
Examples of arithmetic ufunctions (2)

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 125
Ufuncs for rounding decimals
§ Truncation: np.trunc() or np.fix()
§ Rounding: np.round()
§ Floor: np.floor()
§ Ceiling: np.ceil()

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 126
Examples of rounding decimal operations

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 127
Ufuncs for logs
§ Log at base 2: np.log2()
§ Log at base 10: np.log10()
§ Natural log or log at base e: np.log()
§ Log at any base: creating your own ufunc

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 128
Some unary ufuncs

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 129
Some
binary
ufuncs

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 130
Array-Oriented Programming with Arrays
– Expressing Conditional Logic as Array Operations
– Mathematical and Statistical Methods
– Methods for Boolean Arrays
– Sorting
– Unique and Other Set Logic

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 131
Expressing conditional logic as array operations

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 132
Expressing conditional logic as array operations using where

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 133
Conditional logic on 2D arrays

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 134
Conditional logic on 2D arrays (2)

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 135
Mathematical and statistical methods

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 136
Methods for boolean arrays

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 137
Two methods: ndarray.any()and ndarray.all()

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 138
Sorting

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 139
Sorting multidimensional arrays using axis=0

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 140
Sorting multidimensional arrays using axis=1

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 141
Sorting using numpy.sort (creating a copy)

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 142
Unique and other set logic

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 143
Unique and other set logic (2)

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 144
Unique and other set logic: other methods

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 145
File Input and Output with Arrays

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 146
Saving and loading NumPy arrays to/from binary files

20 March 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 147
Saving multiple arrays to a file

20 March 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 148
Lecture summary

§ Introduced the NumPy library and explained why NumPy is important to data analytics.
§ Introduced and describe the NumPy multidimensional arrays (ndarray) with
– Different ways to create arrays,
– The standard data types,
– Different array indexing and slicing methods, and
– Various operations on arrays including arithmetic, linear algebra, aggregate, etc,
– Ways to reshape, concatenate, and split arrays.

§ Introduced various universal functions working on NumPy arrays.

§ Introduced how to save and load NumPy arrays to and from files.

March 20, 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 149
Lecture materials

§ Allen B. Downey. Think Python: How to Think Like a Computer Scientist. O’Reilly Media
Inc., 2024.
§ Luciano Ramalho. Fluent Python, O’Reilly Media Inc., 2015.
§ Brett Slatkin. Effective Python: 125 Specific Ways to Write Better Python, Pearson
Education, Inc., 2025.
§ Wes McKinney. Python for Data Analysis: Data Wrangling with Pandas, NumPy, and Jupyter,
O’Reilly Media Inc., 2022.
§ Guido van Rossum, Barry Warsaw, and Alyssa Coghlan. Python Enhancement Proposal 8
(PEP 8) – Style Guide for Python Code, 2001. URL: https://peps.python.org/pep-0008/

20 March 2025 NumPy Basics • Phan Xuân Hiếu • VNU-UET • [email protected] 150