Nptel Notes 2

Regression
 Regression is a supervised learning that build functional relationship

between dependent and independent variables
Property Rent Price Stock Price
Data Science for Engineers- TA Session 2

Regression Example
 House rent price prediction
Model Rent is 10000 INR
• Location
• Size and area
• Transportation
• Furnished Independent Features Dependent Features
• Utilities
• Pet allowed/not
allowed

Regression Types
Univariate Vs Multivariate
• Univariate: One dependent and one independent variable
• Multivariate: Multiple independent and multiple dependent variables
Square House Price Square Bedrooms

Footage (X) (Y) Footage (X1) (X2) House Price (Y)
1500 $250,000 1500 3 $250,000
1800 $280,000 1800 4 $280,000
1200 $220,000 1200 2 $220,000
2000 $320,000 2000 4 $320,000
1350 $240,000 1350 3 $240,000
Univariate Multivariate
Regression Types
Linear Vs Non-linear
• Linear: Relationship is linear between dependent and independent
variables
• Non-linear: Relationship is nonlinear between dependent and
independent variables

Regression Methods
 Linear
• Ordinary Least Squares (OLS) Regression
• Ridge Regression (L2 Regularization)
• Lasso Regression (L1 Regularization)
• Partial Least Square (PLS) Regression
• Principle Component Analysis (PCA)
 Non-linear
• Polynomial Regression
• Neural Network
• Spline Regression
Regression Process

Regression Illustration
We have the dataset of car service
center
It contains number of cars

(independent variable) and Minute for
Minutes
service (dependent variable)
We want to find the best functional

relationship between both variables
which can be given by linear line
Cars

Ordinary Least square (OLS) 𝒚𝒊 = 𝜷 𝟎 + 𝜷 𝟏 𝒙 𝒊
Linear model between 𝒚𝒊 and 𝒙𝒊 , 𝒊 = 𝟏, … , 𝒏
Error in only dependent variable and no error in 𝝐𝒏

independent variable
Minutes
𝝐𝟏 𝝐𝒊 = 𝒚𝒊 − 𝒚𝒊
The sum of square of errors (SSE)
The minimization of SSE gives estimate of B0 and B1

Cars

Testing goodness of fit
𝑅 is one of the measure use to test determine goodness of fit
R calculates the variability in output variable calculated by input variable
The value of R lie between 1(good fit) and 0 (bad fit)

Adjusted R is the modification of R metric to take into account the
number of independent variables

In a linear regression equation, what does the slope (coefficient) represent?
a) The intercept of the regression line
b) The change in the dependent variable for a unit change in the independent variable
c) The average of the dependent variable
d) The variance of the dependent variable
Q) What does the coefficient of determination

(R-squared) measure in a regression model?
a) The accuracy of the model's predictions
b) The proportion of variance explained by the model
c) The bias of the model
d) The standard erro
We have the following data for which we want to calculate the best fit
R studio

We have the following data for which we want to calculate the best fit
Q1) What is the slope (coefficient) of the best-fitting linear regression line for this
dataset?
a) 2.1
b) 1.3
c) 1.7
d) 2.3
Q2) What is the intercept of the best-fitting linear regression line for this dataset?
a) 2.8
b) 1.8
c) 1.9
d) 2.9
What is the predicted value of Y when X = 6 using the linear regression model?
a) 10.1
b) 8.6
c) 11.3
d) 10.7
What is the (R-squared) for the linear regression model fitted to this dataset?
a) 0.55
b) 0.45
c) 0.35
d) 0.60
What is the mean squared error (MSE) for the linear regression model fitted to this
dataset?
a)3.49
b) 2.93
c) 3.98
d) 4.21
Q) If the slope of the linear regression line is 3 and the intercept is 2, what would be
the predicted Y value when X = 8?
a) 24
b) 26
c) 28
d) 30
Sum Square Quantity Definitions
• SSR (residual sum-of-squares)

• SST (total sum-of-squares)
• SSE (sum-squared error)
• SST = SSE+SSR
• R² = 1-SSE / SST

Which of the following formulas correctly calculates SSE (Sum of Squares Error)?
A) SSE = Σ(yᵢ- ȳ)²
B) SSE = Σ(yᵢ- ŷᵢ)²
C) SSE = Σ(ŷᵢ- ȳ)²
Q) Which equation relates SST, SSE, and SSR?

A) SST = SSE + SSR
B) SST = SSE - SSR
C) SST = SSE * SSR
D) SSE = SST - SSR
If the linear regression model perfectly fits the data, what would be the value of SSE?
A) 0
B) Equal to SST
C) Equal to SSR
D) Indeterminate
Q) What happens to R² when the regression model's fit improves?

A) R² decreases
B) R² increases
C) R² remains unchanged
D) R² becomes negative
Hypothesis test on regression coefficient

Next step to check the linear fit

Residual plots

Nptel Notes 2

Uploaded by

Copyright:

Available Formats

Nptel Notes 2

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Nptel Notes 2

Uploaded by

Copyright:

Available Formats

Regression

 Regression is a supervised learning that build functional relationship

Property Rent Price Stock Price

Data Science for Engineers- TA Session 2

Model Rent is 10000 INR

Data Science for Engineers- TA Session 3

Square House Price Square Bedrooms

Data Science for Engineers- TA Session 5

Data Science for Engineers- TA Session 7

It contains number of cars

We want to find the best functional

Data Science for Engineers- TA Session 8

Error in only dependent variable and no error in 𝝐𝒏

The sum of square of errors (SSE)

The minimization of SSE gives estimate of B0 and B1

Data Science for Engineers- TA Session 9

The value of R lie between 1(good fit) and 0 (bad fit)

Data Science for Engineers- TA Session 10

Q) What does the coefficient of determination

Data Science for Engineers- TA Session 14

• SSR (residual sum-of-squares)

Data Science for Engineers- TA Session 18

Q) Which equation relates SST, SSE, and SSR?

Q) What happens to R² when the regression model's fit improves?

Data Science for Engineers- TA Session 23

Data Science for Engineers- TA Session 24

Data Science for Engineers- TA Session 25

You might also like