CENG450L: Scripintg Lab

Lecture 4
NumPy
Source 3: Python Data Science Handbook
Based on slides prepared By: Dr. Zaher Merhi

1
What is NumpPy?
• NumPy is a Python library used for working with multidimensional
arrays.
• In Python, we have lists that serve the purpose of arrays, but they are
slow to process.
• NumPy provides an efficient multidimensional array object, called
ndarray, providing fast array-oriented arithmetic operations.
• NumPy provides mathematical functions for fast operations on entire
arrays of data without having to write loops.

2
What is NumpPy?
• It has functions for working in domain of linear algebra, Fourier
transform, and random number generation.
• Tools for reading/writing array data to disk and working with memory-
mapped files.
• It provides a C API for connecting NumPy with libraries written in C,
C++, or FORTRAN.
• NumPy is designed for efficiently handling large arrays of data.
• Arrays are very frequently used in data science, signal processing and
image processing, where speed and resources are very important.

3
NumPy array
• NumPy’s array class is called ndarray. It is also known by the alias
array.

4
NumPy array
The more important attributes of an ndarray object are:
• ndarray.ndim: the number of axes (dimensions) of the array.
• ndarray.shape: the dimensions of the array. This is a tuple of integers
indicating the size of the array in each dimension.
For a matrix with n rows and m columns, shape will be (n,m). The
length of the shape tuple is therefore the number of axes, ndim.
• ndarray.dtype: an object describing the type of the elements in the
array.

5
NumPy array
• 1D array

6
NumPy array
• 2D array

7
NumPy array
• 3D array

8
 NumPy vs list: processing speed

• Let’s multiply the sequences by 2 and check the time.

• NumPy is 10 to 100 times faster than pure python and uses significantly
less memory

9
NumPy ndarry: A multidimensional array object
• Generating 2D random arrays

• I can multiply each element by 10

• Or do element by element addition

10
NumPy ndarry: A multidimensional array object
• Nested equal length sequences, like lists, can be converter into
multidimensional arrays:

• arange is an array valued version of the built in Python range function

11
NumPy ndarry: A multidimensional array object

12
Array creation

13
Array Data Types
• You can specify the data type when creating an array:

• Casting from one type to another:

14
Array Data Types
• NumPy data types

15
Element-wise Arithmetic with NumPy Arrays

16
Element-wise Arithmetic with NumPy Arrays

17
Element-wise Arithmetic with NumPy Arrays

18
Basic Indexing and Slicing

Changing values in a slice will lead to the values being

reflected in the original array.

19
Basic Indexing and Slicing
• From 2D to 1 D

Returns row with index 2 (3rd row)

• 2 D indexing

20
Basic Indexing and Slicing
• 3 D Arrays

21
Index and Slicing

22
Index and Slicing

23
Fancy Indexing
• Indexing using integer arrays

24
Fancy Indexing
• Using Negative indices select rows form the end

25
Fancy Indexing
• Indexing using two lists:

26
NumPy functions
• Transpose

27
 NumPy functions
• 2D matrix multiplication: it can be implemented using np.dot() or np.matmul()

28
NumPy functions
• Matrix multiplication

29
NumPy functions
• Element-wise math functions

30
NumPy functions

Element-wise
math functions

31
NumPy functions
• Element-wise maximum: Comparing two arrays

32
33
Expressing conditional Logic as Array Operations

34
Statistics
• Statistical functions are available in NumPy

35
Statistics
• We can specify along which axis the computation should be done.

Along the rows (avg of each column)

Along the columns (avg of each row)

36
Statistics
• Cumulative summation

37
Statistics
• Some of the statistical functions:

38
Linear Algebra
• NumPy package contains numpy.linalg module that provides most the
functionality required for linear algebra:
• Matrix inversion
• Solving systems of linear equation
• Computing eigenvectors and eigenvalues
• Computing determinant
•…

39
Linear Algebra
• Matrix inversion

40
Linear Algebra
• Eigenvalues and eigenvectors

41
Linear Algebra

42
Pseudorandom Generators
• Uniform distribution over [0.0, 1.0[

• Normal Distribution

43
Pseudorandom Generators
• Changing seed

• To avoid a global state use RandomState to create a random number

isolated from others

44