Data Analysis with Python (For Beginners)

reachus@[Link]
 About CloudxLab

Making learning fun and for life

Videos Quizzes Hands-On Projects Case Studies

Real Life Use Cases

 CloudxLab - Playground with
Feedback
Playground for hands-on. System evaluates the code automatically
and nudges the user by giving appropriate feedback

Content Playground

Feedback
CloudxLab - Online Cloud Based Lab

Cloud-based Lab with pre-installed tools and software for

practicing AI, Machine Learning, Deep Learning, Data Science, Big
Data and related technologies
 CloudxLab - Online Cloud Based Lab

Real-world Experience Seamless Experience

Lab setup is exactly same as of setup in No endless downloading/ installations. No
Enterprises. Become job ready from hardware, permissions or configuration
Day 1 issues

Central Dataset Any Device Anywhere

Upload your own dataset Connect from ANY browser,
Or use open source datasets available on lab SSH, device or operating system
 CloudxLab - Social
We learn better with peers. Social proof and leaderboard
increases engagement and motivation
 CloudxLab - Hiring Partners
Dedicated Job Portal → Upgrade career, enhance salary & move
jobs by applying to jobs posted by our hiring partners
CloudxLab - University Partners
 Instructors / Authors

Praveen
Sandeep Giri Abhinav Singh
Pavithran
Founder at [Link] | AI CTO/Co-Founder at Yatis | IOT, Co-Founder, [Link] | AI,
Advisor at Algoworks | Speaker - ML, Computer Vision, Edge ML & Big Data | Visiting Faculty at
AI, Machine Learning, Deep SCMHRD
Learning,Big Data Cypress Semiconductors, Philips,
Multiple patents Byjus, HashCube
Amazon, InMobi, [Link] conference papers, 9+ Years of Exp. in EdTech, Game
18+ Years of Exp. in Enterprise IIT Bombay Dual Degree Development & Building Product
Softwares, Machine Learning &
Churning Humongous Data
What is Python

reachus@[Link]
What is Python

- Python is a interpreted,
high-level language

reachus@[Link]
What is Python

- Python is a interpreted,
high-level language
- Invented in 1991 by Guido van
Rossum

reachus@[Link]
What is Python

- Python is a interpreted,
high-level language
- Invented in 1991 by Guido van
Rossum
- It is easy to use and improves
engineer productivity

reachus@[Link]
What is Python

- Python is a interpreted,
high-level language
- Invented in 1991 by Guido van
Rossum
- It is easy to use and improves
engineer productivity
- Libraries for multiple
applications

reachus@[Link]
What is Python

- Python is a interpreted,
high-level language
- Invented in 1991 by Guido van
Rossum
- It is easy to use and improves
engineer productivity
- Libraries for multiple
applications
- Django framework for web
applications
- We will focus on libraries for
Data Analysis
reachus@[Link]
What is Python

- Python is a interpreted,
high-level language
- Invented in 1991 by Guido van
Rossum
- It is easy to use and improves
engineer productivity
- Libraries for multiple
applications
- Django framework for web
applications
- We will focus on libraries for
Data Analysis
reachus@[Link]
Numpy

reachus@[Link]
What is NumPy

Stands for "Numeric Python" or "Numerical Python".

● Open Source
● Module of Python
● Provides fast mathematical functions

reachus@[Link]
What is NumPy

scikitlearn tensorflow

numpy
Python
matplotlib
pandas

The complete Machine Learning eco-system.

reachus@[Link]
Why use NumPy ?

● Array-oriented computing
● Efficiently implemented multi-dimensional arrays
● Designed for scientific computation
● Library of high-level mathematical functions

reachus@[Link]
Numpy - Introduction

● NumPy’s main object is the homogeneous multidimensional

array
● It is a table of elements
○ usually numbers
○ all of the same type
○ indexed by a tuple of positive integers
● In NumPy dimensions are called axes
● The number of axes is rank

reachus@[Link]
Numpy - Introduction

First Dimension / Axis, Len = 4

Second Dimension / Axis, Len = 3

[[ 0., 0., 0., 0.],

[ 0., 0., 0., 0.],

[ 0., 0., 0., 0.]])

The above array has a rank of 2 since it is 2

dimensional.

reachus@[Link]
Creating Numpy arrays
[Link] - Creating NumPy array from Python Lists/Tuple

Numpy arrays can be created from Python lists or tuple in the

following way.

>>> import numpy as np

>>> a = [Link]([1, 2, 3])
>>> type(a)
<type '[Link]'>
>>> b = [Link]((3, 4, 5))
>>> type(b)
<type '[Link]'>

reachus@[Link]
Creating Numpy arrays
[Link] - An array with all Zeroes

To create an array with all zeroes the function [Link] is

used

>>> x = [Link]( (3,4) )

>>> x
array([[ 0., 0., 0., 0.],
[ 0., 0., 0., 0.],
[ 0., 0., 0., 0.]])

reachus@[Link]
Creating Numpy arrays
[Link] - An array with all Ones

To create an array with all ones the function [Link] is used.

>>> [Link]( (3,4), dtype=np.int16 )

array([[ 1, 1, 1, 1],
[ 1, 1, 1, 1],
[ 1, 1, 1, 1]])

reachus@[Link]
Creating Numpy arrays
[Link] - An array with a given value

To create an array with a given shape and a given value [Link]

is used.

>>> [Link]( (3,4), 0.11 )

array([[ 0.11, 0.11, 0.11, 0.11],
[ 0.11, 0.11, 0.11, 0.11],
[ 0.11, 0.11, 0.11, 0.11]])

reachus@[Link]
Creating Numpy arrays
[Link] - Creating sequence of Numbers

>>> [Link]( 10, 30, 5 )

array([10, 15, 20, 25])
>>> [Link]( 0, 2, 0.3 )
# it accepts float arguments
array([ 0. , 0.3, 0.6, 0.9, 1.2, 1.5, 1.8])

reachus@[Link]
Creating Numpy arrays
[Link] - Creating an array with evenly distributed numbers

● Returns an array having a specific number of points

● Evenly distributed between two values
● The maximum value is included, contrary to arange
Ending Number Total Number of points
Starting Number

>>> [Link](0, 5/3, 6)

array([0. , 0.33333333 , 0.66666667 , 1. , 1.33333333 1.66666667])

reachus@[Link]
Creating Numpy arrays
[Link] - Creating an array with random numbers

Make a 2x3 matrix having random floats between 0 and 1:

>>> [Link](2,3)
array([[ 0.55365951, 0.60150511, 0.36113117],
[ 0.5388662 , 0.06929014, 0.07908068]])

reachus@[Link]
Creating Numpy arrays
[Link] - Creating an empty array

To create an uninitialised array with a given shape. Its content

is not predictable.

>>> [Link]((2,3))
array([[ 0.21288689, 0.20662218, 0.78018623],
[ 0.35294004, 0.07347101, 0.54552084]])

reachus@[Link]
Important attributes of a NumPy object

The NumPy’s array class is called ndarray. The important

attributes of a ndarray object are -

[Link]
the number of axes (dimensions) of the array.
[[ 1., 0., 0.],
[ 0., 1., 2.]]

For the above array the value of [Link] is 2.

reachus@[Link]
Important attributes of a NumPy object

[Link]
the dimensions of the array. This is a tuple of integers
indicating the size of the array in each dimension.
[[ 1., 0., 0.],
[ 0., 1., 2.]]
For the above array the value of [Link] is (2,3)

reachus@[Link]
Important attributes of a NumPy object

[Link]
the total number of elements of the array. This is equal to
the product of the elements of shape.
[[ 1., 0., 0.],
[ 0., 1., 2.]]

For the above array the value of [Link] is 6.

reachus@[Link]
Important attributes of a NumPy object

[Link]
Tells the datatype of the elements in the numpy array. All
the elements in a numpy array have the same type.
>>> c = [Link](1, 5)
>>> [Link]
dtype('int64')

reachus@[Link]
Important attributes of a NumPy object

[Link]
The itemsize attribute returns the size (in bytes) of each
item:
>>> c = [Link](1, 5)
>>> [Link]
8

reachus@[Link]
Reshaping Arrays

The function reshape is used to reshape the numpy array.

The following example illustrates this.

>>> a = [Link](6)
>>> print(a)
[0 1 2 3 4 5]
>>> b = [Link](2, 3)
>>> print(b)
[[0 1 2],
[3 4 5]]

reachus@[Link]
Indexing and Accessing NumPy arrays

reachus@[Link]
Indexing one dimensional NumPy Arrays

0 1 2 3 4 5 6 Index

>>> a = [Link]([1, 5, 3, 19, 13, 7, 3])

>>> a[3]
19
>>> a[2:5] #range
array([ 3, 19, 13])
>>> a[2::2] # How many to jump
array([ 3, 13, 3])
>>> a[::-1] #Go reverse
array([ 3, 7, 13, 19, 3, 5, 1])

reachus@[Link]
Difference with regular Python arrays

1. If you assign a single value to an ndarray slice, it is copied

across the whole slice :
>>> a = [Link]([1, 2, 5, 7, 8])
>>> a[1:3] = -1
>>> a
array([ 1, -1, -1, 7, 8])
----
>>> b = [1, 2, 5, 7, 8]
>>> b[1:3] = -1
TypeError: can only assign an iterable

reachus@[Link]
Difference with regular Python arrays

2. ndarray slices are actually views on the same data buffer. If

you modify it, it is going to modify the original ndarray as well.

>>> a = [Link]([1, 2, 5, 7, 8])

>>> a_slice = a[1:5]
>>> a_slice[1] = 1000
>>> a
array([ 1, 2, 1000, 7, 8])
# Original array was modified

reachus@[Link]
Important attributes of a NumPy object

3. If you want a copy of the data, you need to use the copy
method as another_slice = a[2:6].copy() ,
if we modify another_slice, a remains same.

reachus@[Link]
Indexing multi dimensional NumPy arrays
Multi-dimensional arrays can be accessed as
>>> b[1, 2] # row 1, col 2
>>> b[1, :] # row 1, all columns
>>> b[:, 1] # all rows, column 1

The following format is used while indexing multi-dimensional

arrays
Array[row_start_index:row_end_index, column_start_index:
column_end_index]

reachus@[Link]
Boolean Indexing

We can also index arrays using an ndarray of boolean values on

one axis to specify the indices that we want to access.

>>> a = [Link](12).reshape(3, 4)
>>> rows_on = [Link]([ True, False, True])
>>> a[rows_on , : ] # Rows 0 and 3, all columns
array([[ 0, 1, 2, 3],
[ 8, 9, 10, 11]])

reachus@[Link]
Linear Algebra with NumPy

reachus@[Link]
Vectors

● A vector is a quantity defined by a magnitude and a direction.

● A vector can be represented by an array of numbers called
scalars.

reachus@[Link]
Vectors

For example, say the rocket is going up at a slight angle: it has a

vertical speed of 5,000 m/s, and also a slight speed towards the
East at 10 m/s, and a slight speed towards the North at 50 m/s.
The rocket's velocity may be represented by the following
vector:

velocity 50 m/s

10 m/s

5,000 m/s
reachus@[Link]
Use of Vectors in Machine Learning
● Vectors have many purposes in Machine Learning, most
notably to represent observations and predictions.
● For example, say we built a Machine Learning system to
classify videos into 3 categories (good, spam, clickbait) based
on what we know about them.
Good

Spam

Clickbait

reachus@[Link]
Use of Vectors in Machine Learning
● For each video, we would have a vector representing what
we know about it, such as:

Video

● This vector could represent a video that lasts 10.5 minutes,

but only 5.2% viewers watch for more than a minute, it gets
3.25 views per day on average, and it was flagged 7 times as
spam. As you can see, each axis may have a different
meaning.

reachus@[Link]
Use of Vectors in Machine Learning

● Based on this vector our Machine Learning system may

predict that there is an 80% probability that it is a spam
video, 18% that it is clickbait, and 2% that it is a good video.
This could be represented as the following vector:
Spam

class_probabilities Clickbait
Good

reachus@[Link]
Representing Vectors in Python

● In python, a vector can be represented in many ways, the

simplest being a regular python list of numbers.
○ [1,1,1,1]
● Since Machine Learning requires lots of scientific calculations,
it is much better to use NumPy's ndarray, which provides a
lot of convenient and optimized implementations of essential
mathematical operations on vectors.
● [Link]([1,1,1,1])

reachus@[Link]
Vectorized Operations

● Vectorized operations are far more efficient

● Than loops written in Python to do the same thing
● Let’s test it

reachus@[Link]
Vectorized Operations

Matrix multiplication
1. Using for loop
>>> def multiply_loops(A, B):
C = [Link](([Link][0], [Link][1]))
for i in range([Link][1]):
for j in range([Link][0]):
C[i, j] = A[i, j] * B[j, i]
return C

2. Using NumPy's matrix-matrix multiplication operator

>>> def multiply_vector(A, B):
return A @ B

reachus@[Link]
Vectorized Operations

Matrix multiplication - Sample data

# Two randomly-generated, 100x100 matrices

>>> X = [Link]((100, 100))

>>> Y = [Link]((100, 100))

reachus@[Link]
Vectorized Operations
Matrix multiplication - Loops - timeit Matrix multiplication - Vector - timeit

# First, using the explicit # Second, the NumPy

loops: multiplication:
>>> %timeit >>> %timeit
multiply_loops(X, Y) multiply_vector(X, Y)

4.23 ms ± 107 µs per loop 46.6 µs ± 346 ns per loop

(mean ± std. dev. of 7 runs, (mean ± std. dev. of 7 runs,
100 loops each) 10000 loops each)

Result - It took about 4.23 Result - 46.6 microseconds (46.4

milliseconds (4.23∗10−3 seconds) to ∗10−6 seconds) per multiplication
perform one matrix-matrix
multiplication Conclusion - Two orders of
magnitude faster

reachus@[Link]
Basic Operations on NumPy arrays

reachus@[Link]
Addition in NumPy arrays

Addition can be performed on NumPy arrays as shown below.

They apply element wise.

>>> a = [Link]( [20, 30, 40, 50] )

>>> b = [Link]( 4 )
>>> b
array([0, 1, 2, 3])
>>> c = a + b
>>> c
array([20, 31, 42, 53])

reachus@[Link]
Subtraction in NumPy arrays

Subtraction can be performed on NumPy arrays as shown

below. They apply element wise.
>>> a = [Link]( [20, 30, 40, 50] )
>>> b = [Link]( 4 )
>>> b
array([0, 1, 2, 3])
>>> c = a - b
>>> c
array([20, 29, 38, 47])

reachus@[Link]
Element wise product in NumPy arrays

Element wise product can be performed on NumPy arrays as

shown below.
>>> A = [Link]( [[1,1],
... [0,1]] )
>>> B = [Link]( [[2,0],
... [3,4]] )
>>> A*B # element wise product
array([[2, 0],
[0, 4]])

reachus@[Link]
Matrix Product in NumPy arrays

Matrix product can be performed on NumPy arrays as shown

below.
>>> A = [Link]( [[1,1],
... [0,1]] )
>>> B = [Link]( [[2,0],
... [3,4]] )
>>> [Link](A, B) # matrix product
array([[5, 4],
[3, 4]])

reachus@[Link]
Division in NumPy arrays

Division can be performed on NumPy arrays as shown below.

They apply element wise.

a = [Link]( [20, 30, 40, 50] )

b = [Link](1, 5)
c = a / b
c
array([ 20. , 15. , 13.33333333, 12.5
])

reachus@[Link]
Integer Division in NumPy arrays

Division can be performed on NumPy arrays as shown below.

They apply element wise.

a = [Link]( [20, 30, 40, 50] )

b = [Link](1, 5)
c = a // b
c
array([20, 15, 13, 12])

reachus@[Link]
Modulus in NumPy arrays

Modulus operator can be applied on NumPy arrays as shown

below. They apply element wise.
a = [Link]( [20, 30, 40, 50] )
b = [Link](1, 5)
c = a % b
c
array([0, 0, 1, 2])

reachus@[Link]
Exponents in NumPy arrays

We can find the exponent of each element in a NumPy array

in the following way. It is applied element wise.

a = [Link]( [20, 30, 40, 50] )

b = [Link](1, 5)
c = a ** b
c
array([ 20, 900, 64000, 6250000])

reachus@[Link]
Conditional Operators on NumPy arrays

Conditional operators are also applied element-wise

m = [Link]([20, -5, 30, 40])
m < [15, 16, 35, 36]
array([False, True, True, False], dtype=bool)

m < 25
array([ True, True, False, False], dtype=bool)

To get the elements below 25

m[m < 25]
array([20, -5])

reachus@[Link]
Broadcasting in NumPy arrays

reachus@[Link]
What is Broadcasting ?

1 2 0 2 1 4

4 5 3 4 7 9

1 2 5
???
4 5 7

reachus@[Link]
What is Broadcasting ?

In general, when NumPy expects arrays of the same shape but

finds that this is not the case, it applies the so-called
broadcasting rules.

Basically there are 2 rules of Broadcasting to remember.

reachus@[Link]
First rule of Broadcasting

[[[1, 3 ]]] + [5] [[[6, 8]]]

Shape (1, 1, 2) (1, ) (1, 1, 2)

If the arrays do not have the same rank, then a 1 will be

prepended to the smaller ranking arrays until their ranks match.

reachus@[Link]
First rule of Broadcasting

>>> h = [Link](5).reshape(1, 1, 5)
h
>>> array([[[0, 1, 2, 3, 4]]])
Let's try to add a 1D array of shape (5,) to this 3D array of
shape (1,1,5), applying the first rule of broadcasting.
h + [10, 20, 30, 40, 50] # same as: h + [[[10, 20, 30, 40, 50]]]
array([[[10, 21, 32, 43, 54]]])

reachus@[Link]
Second rule of Broadcasting

reachus@[Link]
Second rule of Broadcasting

On adding a 2D array of shape (2,1) to a 2D ndarray of shape

(2, 3). NumPy will apply the second rule of broadcasting

>>> k = [Link](6).reshape(2, 3)
>>> k
array([[0, 1, 2],
[3, 4, 5]])

>>> k + [100, 200, 300]

array([[100, 201, 302],
[103, 204, 305]])

reachus@[Link]
Mathematical and statistical
functions on NumPy arrays

reachus@[Link]
Finding Mean of NumPy array elements

The ndarray object has a method mean() which finds the mean
of all the elements in the array regardless of the shape of the
numpy array.

>>> a = [Link]([[-2.5, 3.1, 7], [10, 11, 12]])

>>> print("mean =", [Link]())
mean = 6.76666666667

reachus@[Link]
Other useful ndarray methods

Similar to mean there are other ndarray methods which can be

used for various computations.

min - returns the minimum element in the ndarray

max - returns the maximum element in the ndarray
sum - returns the sum of the elements in the ndarray
prod - returns the product of the elements in the ndarray
std - returns the standard deviation of the elements in the
ndarray.
var - returns the variance of the elements in the ndarray.

reachus@[Link]
Other useful ndarray methods
>>> a = [Link]([[-2.5, 3.1, 7], [10, 11, 12]])

>>> for func in ([Link], [Link], [Link], [Link], [Link],

[Link]):
print(func.__name__, "=", func())

min = -2.5
max = 12.0
sum = 40.6
prod = -71610.0
std = 5.08483584352
var = 25.8555555556
reachus@[Link]
Summing across different axes
We can sum across different axes of a numpy array by
specifying the axis parameter of the sum function.

>>> c=[Link](24).reshape(2,3,4)
>>> c
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]]])

reachus@[Link]
Summing across different axes

>>> [Link](axis=0) # sum across matrices

array([[12, 14, 16, 18],
[20, 22, 24, 26],
[28, 30, 32, 34]])

reachus@[Link]
Transposing Matrices
The T attribute is equivalent to calling transpose() when the
rank is ≥2

>>> m1 = [Link](6).reshape(2,3)
>>> m1
array([[0, 1, 2],
[3, 4, 5]])
>>> m1.T
array([[0, 3],
[1, 4],
[2, 5]])

reachus@[Link]
Solving a system of linear scalar equations
The solve function solves a system of linear scalar equations,
such as:

2x + 6y = 6
5x + 3y = -9

reachus@[Link]
Solving a system of linear scalar equations
>>> coeffs = [Link]([[2, 6], [5, 3]])
>>> depvars = [Link]([6, -9])
>>> solution = [Link](coeffs, depvars)
>>> solution
array([-3., 2.])

reachus@[Link]
Solving a system of linear scalar equations
Let’s check the solution.

>>> [Link](solution), depvars

(array([ 6., -9.]), array([ 6, -9]))

reachus@[Link]
References

● NumPy
○ [Link]

reachus@[Link]
 Questions?
[Link]
reachus@[Link]
Pandas

reachus@[Link]
What is Pandas?

● One of the most widely used Python libraries in Data Science after
NumPy and Matplotlib
● The Pandas library Provides
○ High-performance
○ Easy-to-use data structures and
○ Data analysis tools

reachus@[Link]
Pandas - DataFrame

● The main data structure is the DataFrame

● In memory 2D table

○ Like Spreadsheet with column names and row label

reachus@[Link]
Pandas - Data Analysis

● Many features available in Excel are available programmatically like

○ Creating pivot tables

○ Computing columns based on other columns

○ Plotting graphs

reachus@[Link]
Pandas - Data Structures

● Series objects

○ 1D array, similar to a column in a spreadsheet

● DataFrame objects

○ 2D table, similar to a spreadsheet

● Panel objects

○ Dictionary of DataFrames

reachus@[Link]
Pandas - Series Objects

Creating a Series
>>> import pandas as pd
>>> s = [Link]([2,-1,3,5])

Output -
0 2
1 -1
2 3
3 5
dtype: int64

reachus@[Link]
Pandas - Series Objects

Pass as parameters to NumPy functions

>>> import numpy as np
>>> [Link](s)

Output -
0 4
1 1
2 9
3 25
dtype: int64

reachus@[Link]
Pandas - Series Objects

Arithmetic operation on the series

>>> s + [1000,2000,3000,4000]

Output -
0 1002
1 1999
2 3003
3 4005
dtype: int64

reachus@[Link]
Pandas - Series Objects

Broadcasting
>>> s + 1000

Output -
0 1002
1 999
2 1003
3 1005
dtype: int64

reachus@[Link]
Pandas - Series Objects

Binary and conditional operations

>>> s < 0

Output -
0 False
1 True
2 False
3 False
dtype: bool

reachus@[Link]
Pandas - Series Objects

Index labels - Integer location

>>> s2 = [Link]([68, 83, 112, 68])
>>> print(s2)

Output -
0 68
1 83
2 112
3 68
dtype: int64

reachus@[Link]
Pandas - Series Objects

Index labels - Set Manually

>>> s2 = [Link]([68, 83, 112, 68],
index=["alice", "bob", "charles", "darwin"])
>>> print(s2)

Output -
alice 68
bob 83
charles 112
darwin 68
dtype: int64

reachus@[Link]
Pandas - Series Objects

Access the items in series

● By specifying integer location

>>> s2[1]

● By specifying label

>>> s2["bob"]

reachus@[Link]
Pandas - Series Objects

Access the items in series - Recommendations

● Use the loc attribute when accessing by label

>>> [Link]["bob"]

● Use iloc attribute when accessing by integer location

>>> [Link][1]

reachus@[Link]
Pandas - Series Objects

Init from Python dict

>>> weights = {"alice": 68, "bob": 83, "colin": 86,

"darwin": 68}
>>> s3 = [Link](weights)
>>> print(s3)

Output -
alice 68
bob 83
colin 86
darwin 68
dtype: int64
reachus@[Link]
Pandas - Series Objects

Control the elements to include and specify their order

>>> s4 = [Link](weights, index = ["colin", "alice"])

>>> print(s4)

Output -
colin 86
alice 68
dtype: int64

reachus@[Link]
Pandas - Series Objects

Automatic alignment

● When an operation involves multiple Series objects

● Pandas automatically aligns items by matching index labels

reachus@[Link]
Pandas - Series Objects

Automatic alignment - example

>>> print(s2+s3)
Output -
alice 136.0
bob 166.0
charles NaN
colin NaN
darwin 136.0
dtype: float64

* Note NaN

reachus@[Link]
Pandas - Series Objects

Automatic alignment

Do not forget to set the right index labels, else you may get surprising
results
>>> s5 = [Link]([1000,1000,1000,1000])
>>> print(s2 + s5)
Output-
alice NaN
bob NaN
charles NaN
darwin NaN
0 NaN
1 NaN
reachus@[Link]
Pandas - Series Objects

Init with a scalar

>>> meaning = [Link](42, ["life", "universe",

"everything"])
>>> print(meaning)

Output-

life 42
universe 42
everything 42
dtype: int64

reachus@[Link]
Pandas - Series Objects

Series name - A Series can have a name

>>> s6 = [Link]([83, 68], index=["bob", "alice"],

name="weights")
>>> print(s6)

* Here series name is weights

Output-
bob 83
alice 68
Name: weights, dtype: int64

reachus@[Link]
Pandas - Series Objects

Plotting a series

>>> %matplotlib inline

>>> import [Link] as plt
>>> temperatures =
[4.4,5.1,6.1,6.2,6.1,6.1,5.7,5.2,4.7,4.1,3.9,3.5]
>>> s7 = [Link](temperatures, name="Temperature")
>>> [Link]()
>>> [Link]()

reachus@[Link]
Pandas - DataFrame Objects

● A DataFrame object represents

○ A spreadsheet,
○ With cell values,
○ Column names
○ And row index labels

● Visualize DataFrame as dictionaries of Series

reachus@[Link]
Pandas - DataFrame Objects

Creating a DataFrame - Pass a dictionary of Series objects

>>> people_dict = {
"weight": [Link]([68, 83, 112],index=["alice",
"bob", "charles"]),

"birthyear": [Link]([1984, 1985, 1992],

index=["bob", "alice", "charles"], name="year"),

"children": [Link]([0, 3], index=["charles",

"bob"]),

"hobby": [Link](["Biking", "Dancing"],

index=["alice", "bob"]),
}
reachus@[Link]
Pandas - DataFrame Objects

Creating a DataFrame

>>> people = [Link](people_dict)

>>> people

reachus@[Link]
Pandas - DataFrame Objects

Creating a DataFrame - Important Notes

● The Series were automatically aligned based on their index

● Missing values are represented as NaN
● Series names are ignored (the name "year" was dropped)

reachus@[Link]
Pandas - DataFrame Objects

DataFrame - Access a column

>>> people["birthyear"]

Output -

alice 1985
bob 1984
charles 1992
Name: birthyear, dtype: int64

reachus@[Link]
Pandas - DataFrame Objects

DataFrame - Access the multiple columns

>>> people[["birthyear", "hobby"]]

Output -

reachus@[Link]
Pandas - DataFrame Objects

Creating DataFrame - Include columns and/or rows and

guarantee order

>>> d2 = [Link](
people_dict,
columns=["birthyear", "weight", "height"],
index=["bob", "alice", "eugene"]
)
>>> print(d2)

reachus@[Link]
Pandas - DataFrame Objects

DataFrame - Accessing rows

● Using loc
○ [Link]["charles"]
● Using iloc
○ [Link][2]
Output -
birthyear 1992
children 0
hobby NaN
weight 112
Name: charles, dtype: object
reachus@[Link]
Pandas - DataFrame Objects

DataFrame - Get a slice of rows

>>> [Link][1:3]

Output -

reachus@[Link]
Pandas - DataFrame Objects

DataFrame - Pass a boolean array

>>> people[[Link]([True, False, True])]

Output -

reachus@[Link]
Pandas - DataFrame Objects

DataFrame - Pass boolean expression

>>> people[people["birthyear"] < 1990]

Output -

reachus@[Link]
Pandas - DataFrame Objects

DataFrame - Adding and removing columns

>>> # Adds a new column "age"

>>> people["age"] = 2016 - people["birthyear"]

>>> # Adds another column "over 30"

>>> people["over 30"] = people["age"] > 30

>>> # Removes "birthyear" and "children" columns

>>> birthyears = [Link]("birthyear")
>>> del people["children"]

>>> people

reachus@[Link]
Pandas - DataFrame Objects

DataFrame - A new column must have the same number of rows

>>> # alice is missing, eugene is ignored

>>> people["pets"] = [Link]({

"bob": 0,
"charles": 5,
"eugene":1
})

>>> people

reachus@[Link]
Pandas - DataFrame Objects

DataFrame - Add a new column using insert method after an

existing column

>>> [Link](1, "height", [172, 181, 185])

>>> people

reachus@[Link]
Pandas - DataFrame Objects

DataFrame - Add new columns using assign method

>>> (people
.assign(body_mass_index = lambda df:df["weight"]
/ (df["height"] / 100) ** 2)
.assign(overweight = lambda df:
df["body_mass_index"] > 25)
)

reachus@[Link]
Pandas - DataFrame Objects

DataFrame - Sorting a DataFrame

● Use sort_index method

○ It sorts the rows by their index label
○ In ascending order
○ Reverse the order by passing ascending=False
○ Returns a sorted copy of DataFrame

reachus@[Link]
Pandas - DataFrame Objects

DataFrame - Sorting a DataFrame

>>> people.sort_index(ascending=False)

reachus@[Link]
Pandas - DataFrame Objects

DataFrame - Sorting a DataFrame - inplace argument

>>> people.sort_index(inplace=True)
>>> people

reachus@[Link]
Pandas - DataFrame Objects

DataFrame - Sorting a DataFrame - Sort By Value

>>> people.sort_values(by="age", inplace=True)

>>> people

reachus@[Link]
Pandas - DataFrame Objects

Plotting a DataFrame

>>> [Link](
kind = "line",
x = "body_mass_index",
y = ["height", "weight"]
)
>>> [Link]()

reachus@[Link]
Pandas - DataFrame Objects

DataFrames - Saving and Loading

● Pandas can save DataFrames to various backends such as

○ CSV
○ Excel (requires openpyxl library)
○ JSON
○ HTML
○ SQL database

reachus@[Link]
Pandas - DataFrame Objects

DataFrames - Saving

Let’s create a new DataFrame my_df and save it in various formats

>>> my_df = [Link](

[
["Biking", 68.5, 1985, [Link]],
["Dancing", 83.1, 1984, 3]
],
columns=["hobby","weight","birthyear","children"],
index=["alice", "bob"]
)
>>> my_df

reachus@[Link]
Pandas - DataFrame Objects

DataFrames - Saving

● Save to CSV
○ >>> my_df.to_csv("my_df.csv")
● Save to HTML
○ >>> my_df.to_html("my_df.html")
● Save to JSON
○ >>> my_df.to_json("my_df.json")

reachus@[Link]
Pandas - DataFrame Objects

DataFrames - What was saved?

>>> for filename in ("my_df.csv", "my_df.html",

"my_df.json"):
print("#", filename)
with open(filename, "rt") as f:
print([Link]())
print()

reachus@[Link]
Pandas - DataFrame Objects

DataFrames - What was saved?

Note that the index is saved as the first column (with no name) in a CSV file

reachus@[Link]
Pandas - DataFrame Objects
DataFrames - What was saved?

Note that the index is saved as <th> tags in HTML

reachus@[Link]
Pandas - DataFrame Objects

DataFrames - What was saved?

Note that the index is saved as keys in JSON

reachus@[Link]
Pandas - DataFrame Objects

DataFrames - Loading

● read_csv # For loading CSV files

● read_html # For loading HTML files

● read_excel # For loading Excel files

reachus@[Link]
Pandas - DataFrame Objects

DataFrames - Load CSV file

>>> my_df_loaded = pd.read_csv("my_df.csv", index_col=0)

>>> my_df_loaded

reachus@[Link]
Pandas - DataFrame Objects

DataFrames - Overview

● When dealing with large DataFrames, it is useful to get a quick overview

of its content
● Load [Link] inside dataset directory to create a DataFrame and
get a quick overview

reachus@[Link]
Pandas - DataFrame Objects

DataFrames - Overview

● Let’s understand below methods

○ head()
○ tail()
○ info()
○ describe()

reachus@[Link]
Pandas - DataFrame Objects

DataFrames - Overview - head()

● The head method returns the top 5 rows

>>> housing = pd.read_csv("dataset/[Link]")

>>> [Link]()

reachus@[Link]
Pandas - DataFrame Objects

DataFrames - Overview - tail()

● The tail method returns the bottom 5 rows

● We can also pass the number of rows we want

>>> [Link](n=2)

reachus@[Link]
Pandas - DataFrame Objects

DataFrames - Overview - info()

● The info method prints out the summary of each column's contents

>>> [Link]()

reachus@[Link]
Pandas - DataFrame Objects

DataFrames - Overview - describe()

● The describe method gives a nice overview of the main aggregated

values over each column
○ count: number of non-null (not NaN) values
○ mean: mean of non-null values
○ std: standard deviation of non-null values
○ min: minimum of non-null values
○ 25%, 50%, 75%: 25th, 50th and 75th percentile of non-null values
○ max: maximum of non-null values
reachus@[Link]
References

● Pandas
○ [Link]

reachus@[Link]
 Questions?
[Link]
reachus@[Link]
Matplotlib

reachus@[Link]
Matplotlib - Overview

● Matplotlib is a Python 2D plotting library

● Produces publication quality figures in a variety of
○ Hardcopy formats and
○ Interactive environments

reachus@[Link]
Matplotlib - Overview

● Matplotlib can be used in

○ Python scripts
○ Python and IPython shell
○ Jupyter notebook
○ Web application servers
○ GUI toolkits

reachus@[Link]
Matplotlib - pyplot Module

● [Link]
○ Collection of functions that make matplotlib work like MATLAB
○ Majority of plotting commands in pyplot have MATLAB analogs with
similar arguments

reachus@[Link]
Matplotlib - pyplot Module

● [Link]
○ Collection of functions that make matplotlib work like MATLAB
○ Majority of plotting commands in pyplot have MATLAB analogs with
similar arguments

reachus@[Link]
Matplotlib - pyplot Module - plot()

>>> import [Link] as plt

>>> [Link]([1,2,3,4])
>>> [Link]('some numbers')
>>> [Link]()

reachus@[Link]
Matplotlib - pyplot Module - plot()

plot x versus y
>>> import [Link] as plt
>>> [Link]([1, 2, 3, 4], [1, 4, 9, 16])
>>> [Link]('some numbers')
>>> [Link]()

reachus@[Link]
Matplotlib - pyplot Module - Histogram

>>> import [Link] as plt

>>> x =
[21,22,23,4,5,6,77,8,9,10,31,32,33,34,35,36,37,18,49,50,
100]
>> num_bins = 5
>> [Link](x, num_bins, facecolor='blue')
>> [Link]()

reachus@[Link]
References

● Matplotlib
○ [Link]

reachus@[Link]
 Questions?
[Link]
reachus@[Link]