Machine Learning

Subject Subject Title Credit Lecture Tutorial Practical Type

Code
CORE–Machine Learning 4 4 - - Theory

Goal: For Students to Explain about different Machine Learning Algorithms and Understand about
supervised and Unsupervised Learning Techniques.
Course Objective:

1. To understand the basic concepts of statistical learning methods and models.

2. To understand the importance of supervised learning in multivariate data sets.
3. To understand the estimation procedure for multiple regression coefficients
4. To understand the assumptions in estimating regression coefficients using OLS
method.
5. To understand the importance of supervised learning and unsupervised learning
algorithms for prediction.

Course Outcomes:

CO1 : Understand the difference between continuous class label and discrete class label
classification methods.
CO2 : Predict the continuous class variable using linear regression analysis.
CO3 : Predict the binary class variable using decision tree and random forest.
CO4 : Understand the importance of Logistic regression and its application in business.
CO5 : Understand the important concepts of neural networks and its prediction
techniques.
Module -1 12 Hrs.

Introduction to Machine Learning Algorithms:

Introduction to Machine learning – Statistical Learning – types of Machine Learning –

learning models: geometric, probabilistic and logistic models, introduction to supervised,
unsupervised and reinforcement learning – model evaluation – model implementation –
model accuracy indicators.

Module -2 12 Hrs.

Supervised Learning –Regression Analysis:

Introduction to parametric machine learning method, assumptions of parametric machine learning

methods, linear model and its assumptions, simple linear regression, parameter estimation,
properties of regression parameters, testing the significance of regression parameters, estimation of
σ^2, Interval Estimation of the Mean Response, prediction of new observations, Confidence
interval for β_0, β_1 and σ^2, Multiple linear Regression analysis, parameter estimation, and
significance of coefficients, assumptions of multiple linear regression parameters.
Module - 3 12 Hrs.

Classification Techniques – Decision Tree:

Introduction to decision tree algorithms, classification tree, characteristics of classification tree –
size and hierarchical nature of tree, training and testing data set, induction algorithms, probability
estimation in decision tree – Laplace correction and no match method, stopping criteria for tree
development, pruning techniques and pruned tree, evaluation of decision tree classifiers,
generalization error, F measure, Confusion matrix, ROC curve, Hit Rate Curve, Lift curve,
McNemar’s Test, Resample paired t test, K-fold cross validated paired t test, prediction using better
model, Decision tree ensembles methods.
Module-4 12 Hrs.

Classification Techniques – Logistic Regression:

Introduction to logistic regression, assumptions involved in logistic regression, concepts on odds

and odds ratio, maximum likelihood estimation, binomial logistic regression, parameter estimation,
properties of logistic regression coefficients, logistic regression for correlated data, model accuracy
testing, confusion matrix, Receiver Operating Characteristic Curve, area under curve, likelihood
ratio test, concepts and interpretation of Pseudo R square tests, Hosmer-Lemeshow Test, Wald Test,
prediction using better fit model and interpretation.
Module-5 12 Hrs.

Unsupervised Learning:

Introduction to data dimension reduction techniques, linearity of variables, assumptions of linearity

among variables, general purpose and description of principle component analysis, extraction of
principle components, extraction techniques, orthogonal and oblique rotation of linear combination
of variables, factor analysis and its relevance with business application, introduction to cluster
analysis and its validations.
References:
1. Introduction to Linear Regression Analysis, Fifth Edition - Douglas C. Montgomery,
Elizabeth A. Peck, G. Geoffrey Vining, A John Wiley & Sons, inc., publication
2. Introduction to Machine Learning - EthemAlpaydm, The MIT Press
3. Applied Regression Analysis, Third Edition – Norman R Draper, Harry Smith, John Wiley
& Sons.
4. Using Multivariate Statistics - Barbara G. Tabachnick, Linda S. Fidell, Pearson Education Inc
Mapping of Course Outcomes with Program Outcomes:

Course Program Outcomes

Outcomes P01 P02 P03 P04 P05 P06 P07 P08 P09
1 H H H L L L H H H
2 M H M H M H H L M
3 H M H H M H L H H
4 H H M L H L H H H
5 H H H H H M H H H
Machine Learning Lab
Subject Subject Title Credit Lecture Tutorial Practical Type
Code
Core Practical Machine 2 - - 4 Practica
Learning LAB l

List of Programs:

Exercise – 1

Consider the following table on Air Quality

S.No Ozone Solar R Wind Temp Month Day

1 41 190 7.4 67 5 1

2 36 118 8 72 5 2

3 12 149 12.6 74 5 3

4 18 313 11.5 62 5 4

5 27 192 14.3 56 5 5

6 28 193 14.9 66 5 6

7 23 299 8.6 65 5 7

8 19 99 13.8 59 5 8

9 8 19 20.1 61 5 9

10 24 194 8.6 69 5 10

11 7 152 6.9 74 5 11

12 16 256 9.7 69 5 12

13 11 290 9.2 66 5 13

14 14 274 10.9 68 5 14

15 18 65 13.2 58 5 15

16 14 334 11.5 64 5 16

17 34 307 12 66 5 17

18 6 78 18.4 57 5 18
 19 30 322 11.5 68 5 19

20 11 44 9.7 62 5 20

1. Summarize the above table in R

2. Create the above table in data frame
format in R without importing from outer source.
3. Find the linear regression line on
given table taking ozone as dependent variable.
4. Predict 21st day of ozone level in the
air with given factors.
5. Find the autocorrelation of error
produced from the fitted line
6. Analyse multicollinearity among
independent variables and find the suitable solution to remove multicollinearity.
7. Find the variance among error terms
and comment on the equal variance among error terms in the output.
8. Estimate the presence of
autocorrelation using Durbin – Watson test statistic.

Exercise - 2
1. Estimate appropriate regression line with suitable predictors. Compare different
regression lines and comment on regression coefficients.
2. Estimate the significance of regression coefficients using ANOVA and compare with
F and partial t test.
3. Model fit using R Square and Adjusted R square values.
4. Estimate Cook Statistic and Press Statistic for diagnostic checking
5. Post model statistical testing for the better fit and error free prediction.
6. Normality testing on error terms of fitted model

Exercise - 3
1. Plot residual versus Fitted values using plot command

2. Plot residual versus Observed using Plot command

3. Plot observed versus and fitted values using plot command

4. Find out the leverage value in the fitted values using which.max command.
5. Interpret the residual summary from the lm( ) command.

6. Find out the VIF values using inbuilt function available in R.