Description
Intro to NumPy
NumPy is the fundamental package for scientific computing with Python. It’s used for working with
arrays and contains functions for working in the domain of linear algebra, fourier transform, and
matrices. In Python we have list, which serve the purpose of arrays, so why do we bother learning
NumPy in the first place? Well, NumPy arrays are much faster than traditional Python lists and provide
many supporting functions that make working with arrays easier. Part of why they’re significantly
faster is because the parts that require fast computation are written in C or C++.
Let’s begin by importing NumPy and learning how to create NumPy arrays. If for some reason you
don’t have numpy installed, you will first have to go to your terminal (or Anaconda Prompt if on
Windows) and enter the following:
conda
install
numpy
Make sure you’ve already installed Anaconda
Creating Arrays and Common Methods
The first way of creating a NumPy array is by converting your existing Python list.
[1,
2,
3,
4,
5]
<class
’list’>
[1
2
3
4
5]
<class
’numpy.ndarray’>
[[1,
2,
3,
4,
5],
[6,
7,
8,
9,
10],
[11,
12,
13,
14,
15]]
<class
’list’>
[[
1
2
3
4
5]
[
6
7
8
9
10]
[11
12
13
14
15]]
<class
’numpy.ndarray’>
However, you are more likely to use some of NumPy’s built in methods to generate ndarrays. Here we’ll
introduce you to a few of these built in methods.
arange(start, stop, step) will return evenly spaced values within a given interval. The default step
size is 1.
array([
0,
1,
2,
3,
4,
5,
6,
7,
8,
9,
10,
11,
12,
13,
14])
array([
0,
2,
4,
6,
8,
10,
12,
14])
What if we wanted a 2d array instead? We can call reshape(rows, columns) on an existing NumPy
array. Please note that the product of rows and columns must evaluate to the total number of elements
in your current NumPy array.
array([[
0,
1,
2,
3,
4],
[
5,
6,
7,
8,
9],
[10,
11,
12,
13,
14]])
If we’d like to generate an ndarry of zeroes or ones (useful with certain calculations), we could do so
by simply calling zeros or ones. For example:
array([0.,
0.,
0.,
0.,
0.,
0.,
0.,
0.,
0.,
0.,
0.,
0.,
0.,
0.,
0.])
array([[0.,
0.,
0.,
0.,
0.],
[0.,
0.,
0.,
0.,
0.],
[0.,
0.,
0.,
0.,
0.]])
array([1.,
1.,
1.,
1.,
1.,
1.,
1.,
1.,
1.,
1.,
1.,
1.,
1.,
1.,
1.])
array([[1.,
1.,
1.,
1.,
1.],
[1.,
1.,
1.,
1.,
1.],
[1.,
1.,
1.,
1.,
1.]])
You might have noticed that these values defaulted to floats. If for some reason you’d like to use a
different type, say, int, you can insert an additional paramater such as dtype=int. See more on dtypes
here
array([[1,
1,
1,
1,
1],
[1,
1,
1,
1,
1],
[1,
1,
1,
1,
1]])
A common matrix used in linear algebra is the identity matrix (an n × n square matrix with ones on the
main diagonal and zeros elsewhere). We can generate this in NumPy using eye(n).
array([[1.,
0.,
0.,
0.,
0.],
[0.,
1.,
0.,
0.,
0.],
[0.,
0.,
1.,
0.,
0.],
[0.,
0.,
0.,
1.,
0.],
[0.,
0.,
0.,
0.,
1.]])
Another common use case is to generate an ndarray of random numbers. This can be done in 2 ways.
rand (which will fill the ndarray with random samples from a uniform distribution over [0, 1)), and
randn (which will return a sample (or samples) from the standard normal distribution.) Additionally, we
can use randint(low, high, size) to generate a single or multiple random integers between [low, high).
The size parameter specifies how many we’d like. Let’s see a few examples.
array([0.44538557,
0.43482283,
0.20369315,
0.44795307,
0.30920162])
array([0.4242536
,
0.71231944,
0.63292466,
0.79684565,
0.96175713])
73
array([99,
50,
67,
47,
50,
54,
39,
24,
55,
67,
76,
65,
8,
84,
41])
Other common methods that you’re likely to encounter in this class include min, max, argmin, and
argmax. The only difference between the two arg methods is that they’ll instead return the index
position of the min/max value. For example:
array([[19,
20,
74,
33,
47],
[53,
51,
25,
78,
46],
[
1,
14,
41,
28,
81],
[33,
50,
54,
67,
15]])
The
smallest
value
in
A
is
1,
and
is
a
located
at
position
10
The
largest
value
in
A
is
81,
and
is
a
located
at
position
14
Lastly, we’ll often find ourselves wanting to know the shape (dimensions) of our ndarray. This can be
done using shape.
(4,
5)
(4,
5)
Great! you now know how to create NumPy arrays and some of the common methods. Let’s now look
into some common operations we can perform on these arrays.
Common Operations
A:
[
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15]
B:
[
1
3
5
7
9
11
13
15
17
19
21
23
25
27
29]
C:
[[0
1]
[2
3]]
D:
[[0
1]
[2
3]]
E:
[[
1
2
3
4
5]
[
6
7
8
9
10]
[11
12
13
14
15]]
F:
[
0
1
2
3
4
5
6
7
8
9
10]
Arithmetic
array([
2,
5,
8,
11,
14,
17,
20,
23,
26,
29,
32,
35,
38,
41,
44])
array([
6,
7,
8,
9,
10,
11,
12,
13,
14,
15,
16,
17,
18,
19,
20])
array([
0,
1,
2,
3,
4,
5,
6,
7,
8,
9,
10,
11,
12,
13,
14])
array([4,
3,
2,
1,
0,
1,
2,
3,
4,
5,
6,
7,
8,
9,
10])
array([
1,
6,
15,
28,
45,
66,
91,
120,
153,
190,
231,
276,
325,
378,
435])
array([
5,
10,
15,
20,
25,
30,
35,
40,
45,
50,
55,
60,
65,
70,
75])
array([1.
,
0.66666667,
0.6
,
0.57142857,
0.55555556,
0.54545455,
0.53846154,
0.53333333,
0.52941176,
0.52631579,
0.52380952,
0.52173913,
0.52
,
0.51851852,
0.51724138])
array([0.5,
1.
,
1.5,
2.
,
2.5,
3.
,
3.5,
4.
,
4.5,
5.
,
5.5,
6.
,
6.5,
7.
,
7.5])
Note: Becareful dividing by zero. You’ll get nan short for Not a number.
As you may have noticed, the standard operations *, +, ‑, / work element‑wise on arrays. If you’d like to
instead do matrix multiplication, matmul can be used. Here’s a quick reference in case you forgot how
matrix multiplication works.
array([[
2,
3],
[
6,
11]])
Universal Functions
NumPy also contains universal functions, which is a function that operates on ndarrays in an element‑
by‑element fashion. Let’s see a few examples.
array([[1.
,
1.41421356,
1.73205081,
2.
,
2.23606798],
[2.44948974,
2.64575131,
2.82842712,
3.
,
3.16227766],
[3.31662479,
3.46410162,
3.60555128,
3.74165739,
3.87298335]])
array([[0.
,
0.69314718,
1.09861229,
1.38629436,
1.60943791],
[1.79175947,
1.94591015,
2.07944154,
2.19722458,
2.30258509],
[2.39789527,
2.48490665,
2.56494936,
2.63905733,
2.7080502
]])
Something that will often come in handy is the where(condition, x, y) method. This will loop through
every element in your ndarray and return a new ndarray that replaces the element with x if the
condition is met, and y if the condition is not met. In the example below we’re multiplying all odd
numbers by 100, else replacing with ‑1. Try changing the ‑1 to F in the example below and see what
happens.
array([
1,
100,
1,
300,
1,
500,
1,
700,
1,
900,
1])
Nice! Now that you know how to create NumPy arrays and perform basic operations on them, let’s take
a look at how to index and select certain elements.
Indexing
Indexing 1d array
Note: recall that counting starts from 0. Given an array [5, 6, 7, 8], we say that value 5 is at
index/position 0.
array([
0,
1,
2,
3,
4,
5,
6,
7,
8,
9,
10,
11,
12,
13,
14])
Obtaining a single element will look similar to Python arrays:
A[0]:
0
A[5]:
5
A[14]:
14
We can grab a section using A[start_index : stop_index]. stop_index is not inclusive.
array([0,
1,
2,
3,
4,
5,
6,
7,
8,
9])
We can also modify values in this way.
array([500,
500,
500,
500,
500,
500,
500,
500,
500,
500,
10,
11,
12,
13,
14])
Indexing a 2d array
array([[
0,
1,
2,
3,
4,
5,
6,
7,
8,
9],
[10,
11,
12,
13,
14,
15,
16,
17,
18,
19],
[20,
21,
22,
23,
24,
25,
26,
27,
28,
29],
[30,
31,
32,
33,
34,
35,
36,
37,
38,
39],
[40,
41,
42,
43,
44,
45,
46,
47,
48,
49]])
Obtaining a single element can be done using B[row, col].
0th
row
and
2nd
col
in
B:
2
3rd
row
and
3rd
col
in
B:
33
4th
row
and
2nd
col
in
B:
42
Similar to before, we can also grab a section of interest from this 2darray. Only now we have to specify
both the row sections of interest and column sections of interest. B[02, 35] says: I want rows 0‑2
from B, but only the elements in that row corresponding to columns 3‑5. Recall that stop_index is not
inclusive. Here stop_index for the rows is 2, and stop_index for the columns is 5.
array([[
3,
4],
[13,
14]])
Again, we can also modify values in this way.
array([[
0,
1,
2,
1,
1,
5,
6,
7,
8,
9],
[10,
11,
12,
1,
1,
15,
16,
17,
18,
19],
[20,
21,
22,
23,
24,
25,
26,
27,
28,
29],
[30,
31,
32,
33,
34,
35,
36,
37,
38,
39],
[40,
41,
42,
43,
44,
45,
46,
47,
48,
49]])
Boolean Array Indexing
array([[
0,
1,
2,
3,
4],
[
5,
6,
7,
8,
9],
[10,
11,
12,
13,
14],
[15,
16,
17,
18,
19],
[20,
21,
22,
23,
24],
[25,
26,
27,
28,
29],
[30,
31,
32,
33,
34],
[35,
36,
37,
38,
39],
[40,
41,
42,
43,
44],
[45,
46,
47,
48,
49]])
A really cool feature of NumPy is that we can index arrays using comparison operators. This Lets us
modify or select only elements meeting some condition. To demonstrate this, lets first see what is
returned when we try to use comparison operators with our arrays.
array([[
True,
False,
True,
False,
True],
[False,
True,
False,
True,
False],
[
True,
False,
True,
False,
True],
[False,
True,
False,
True,
False],
[
True,
False,
True,
False,
True],
[False,
True,
False,
True,
False],
[
True,
False,
True,
False,
True],
[False,
True,
False,
True,
False],
[
True,
False,
True,
False,
True],
[False,
True,
False,
True,
False]])
We get back an array of the same shape telling us which values in C satisfy the condition C%2==0
(even values). Recall that 0 is an alias for False, and 1 is an alias for True. Because of this, we can
actually call the sum function right off of this array, which will evaluate to the total number of even
numbers in C.
25
Something we’ll find ourselves doing more often is passing the boolean array as an index. What this
will do is filter out the False elements and only leave us with the elements corresponding to True (even
values in our case).
array([
0,
2,
4,
6,
8,
10,
12,
14,
16,
18,
20,
22,
24,
26,
28,
30,
32,
34,
36,
38,
40,
42,
44,
46,
48])
Congratulations! You now know how to create NumPy arrays, perform common operations on them,
and how to index them. There’s so much that NumPy can do, but this should cover just about all that
you’ll need to succeed in this course. Other popular Python libraries used for data science such as
Pandas and Matplotlib are built on top of NumPy, so you’ll be using a lot of these features alongside
those libraries.
In
[1]:
import numpy as np
In
[2]:
python_list = [1, 2, 3, 4, 5]
print(python_list)
print(type(python_list))
In
[3]:
numpy_array = np.array(python_list)
print(numpy_array)
print(type(numpy_array))
In
[4]:
python_2d_array = [[1, 2, 3, 4, 5], [6, 7, 8, 9, 10], [11, 12, 13, 14, 15]]
print(python_2d_array)
print(type(python_2d_array))
In
[5]:
numpy_2d_array = np.array(python_2d_array)
print(numpy_2d_array)
print(type(numpy_2d_array))
In
[6]:
np.arange(0, 15)
Out[6]: In
[7]:
np.arange(0, 15, 2)
Out[7]: In
[8]:
np.arange(0, 15).reshape(3, 5)
Out[8]: In
[9]:
np.zeros(15)
Out[9]: In
[10]:
np.zeros(15).reshape(3, 5)
Out[10]: In
[11]:
np.ones(15)
Out[11]: In
[12]:
np.ones(15).reshape(3, 5)
Out[12]: In
[13]:
np.ones(15, dtype=int).reshape(3, 5)
Out[13]: In
[14]:
np.eye(5)
Out[14]: In
[15]:
np.random.rand(5)
Out[15]: In
[16]:
np.random.randn(5)
Out[16]: In
[17]:
np.random.randint(1,100)
Out[17]: In
[18]:
np.random.randint(1, 100, 15)
Out[18]: In
[19]:
A = np.random.randint(0, 100, 20).reshape(4, 5)
A
Out[19]: In
[20]:
print(f’The
smallest
value
in
A
is
{A.min()},
and
is
a
located
at
position
{A.argmin()
print(f’The
largest
value
in
A
is
{A.max()},
and
is
a
located
at
position
{A.argmax()}
In
[21]:
A.shape
Out[21]: In
[22]:
np.shape(A)
Out[22]: In
[23]:
A = np.arange(1, 16)
B = np.arange(1, 30, 2)
C = np.arange(0, 4).reshape(2, 2)
D = np.arange(0, 4).reshape(2, 2)
E = np.arange(1, 16).reshape(3, 5)
F = np.arange(11)
print(f’A:
{A}’)
print(f’B:
{B}’)
print(‘C:’)
print(C)
print(‘D:’)
print(D)
print(‘E:’)
print(E)
print(f’F:
{F}’)
In
[24]:
A + B
Out[24]: In
[25]:
A + 5
Out[25]: In
[26]:
A B
Out[26]: In
[27]:
A 5
Out[27]: In
[28]:
A * B
Out[28]: In
[29]:
A * 5
Out[29]: In
[30]:
A / B
Out[30]: In
[31]:
A / 2
Out[31]: In
[32]:
np.matmul(C, D)
Out[32]: In
[33]:
np.sqrt(E)
Out[33]: In
[34]:
np.log(E)
Out[34]: In
[35]:
np.where(F%2==0,
1,
F*100)
Out[35]: In
[36]:
A = np.arange(15)
A
Out[36]: In
[37]:
print(f’A[0]:
{A[0]}’)
print(f’A[5]:
{A[5]}’)
print(f’A[14]:
{A[14]}’)
In
[38]:
A[0:10]
Out[38]: In
[39]:
A[0:10] = 500
A
Out[39]: In
[40]:
B = np.arange(50).reshape(5, 10)
B
Out[40]: In
[41]:
print(f’0th
row
and
2nd
col
in
B:
{B[0, 2]}’)
print(f’3rd
row
and
3rd
col
in
B:
{B[3, 3]}’)
print(f’4th
row
and
2nd
col
in
B:
{B[4, 2]}’)
In
[42]:
B[0:2, 3:5]
Out[42]: In
[43]:
B[0:2, 3:5] = 1
B
Out[43]: In
[44]:
C = np.arange(50).reshape(10, 5)
C
Out[44]: In
[45]:
C%2
==
0
Out[45]: In
[46]:
(C%2
==
0).sum()
Out[46]: In
[47]:
C[C%2
==
0]
Out[47]:
