Shalabh
shalab@iitk.ac.in
shalabh1@yahoo.com
Department of Mathematics & Statistics
Indian Institute of Technology Kanpur, Kanpur - 208016 (India)

HOME PAGE


 

Multivariate Data Mining - Methods and Applications

 

by

Professor Anoop Chaturvedi

 

Allahabad University

Swayam Prabha Course

The forty hours course is for the students in Bachelor's and Master's programmes and covers the topics of data mining

Suggested books:

(i)              Izenman, A.J., (2008), Modern Multivariate Statistical Techniques: Regression, Classification, and Manifold Learning, Springer.

(ii)            James, G., Witten D., Hastie T., Tibshirani R., (2013), An Introduction to Statistical Learning with applications to R, Springer.

(iii)          Everitt B.S., Landau S., Leese M., Stahl D. (2011), Cluster Analysis, 5th Edition, Wiley.

(iv)           Han, J. and Kamber, M (2006). Data Mining: Concepts and Techniques, 2nd edition, Morgan Kaufmann.

(v)             Dunham, M. H. (2003). Data Mining: Introductory and Advanced Topics, Pearson Education.

 

Language of the course: English

 

Duration of the course: 40 Hours

 

Swayam Prabha DTH Channel 16 Youtube link: The telecasted lectures are available at  YouTube (Click here)

 

Slides and Videos used in the lectures: 

 

 

Lecture No.

Lecture videos download links

Lecture slides download links

Brief Description

Lecture Title

1

Lecture 1

L 1

Introduction

Introduction to data mining, its applications in various fields, Outline of the course

2

Lecture 2

L 2

Data Mining, Machine Learning and Artificial Intelligence

Basics of Data mining, Data mining and knowledge discovery, Artificial Intelligence,

3

Lecture 3

L 3

Machine Learning Rules

Machine Learning Rules, Supervised, unsupervised learning, Batch Learning and Online learning, Reinforcement learning, resubstitution Estimate, Generalizations for improving resubstitution estimates, Training, learning and test sets, Bootstrap, Ockham’s (or Occam’s) razor principle, methods for reducing the effects of overfitting, Sampling Design for obtaining data,

4

Lecture 4

L 4

Matrix Algebra

Introduction to vectors, Operations of vectors, Different types of vectors, different types of matrices, matrix operations, Eigen values and eigen vectors, different results related to orthogonal matrices, idempotent matrices, quadratic forms, Matric norms

5

Lecture 5

L 5

Multivariate Analysis

Multivariate probability distributions, Multivariate normal distribution, marginal and conditional distributions, Expectation of some quadratic forms

6

Lecture 6

L 6

Multiple Regression Model: Introduction

General structure of regression problem, Multiple linear models, Estimation of parameters, model in deviation form

7

Lecture 7

L 7

Properties of Estimators and Model Selection Criterion

Properties of estimators and model selection criterion, R square, adjusted R square, AIC, BIC

8

Lecture 8

L 8

Model Assessment for Multiple Regression

Model Assessment for random and fixed X, Prediction error, apparent error rate or resubstitution error rate, resampling methods, V-fold cross validation, Optimism corrected bootstrap estimate of PE

9

Lecture 9

L 9

Multicollinearity and Variables Selection

Multicollinearity problem and its implications and measures, stepwise variable selection regression, backward, and forward methods, hybris stepwise method 

10

Lecture 10

L 10

Shrinkage Estimation

Shrinkage estimation, penalized regression estimators, LASSO and Ridge regression

11

Lecture 11

L 11

Principal Component and Least Angle Regression

Principal Component regression and Least Angle Regression methods

12

Lecture 12

L 12

Regression Methods for Classification

Formulation of probability models, LOGIT and PROBIT Models for classification

13

Lecture 13

L 13

Data Mining Methods for High Dimensional data: Principal Component Analysis

The Curse of Dimensionality, Basics and objectives of Principal Component Analysis for linear feature space, Advantages and Disadvantages

14

Lecture 14

L 14

Statistical Analysis of PCA

Population PCA, Least-Squares Optimality of PCA, Eckart-Young Theorem, Courant–Fischer Min-Max theorem, PCA as a Variance-Maximization Technique

15

Lecture 15

L 15

Sample PCA and Ap16plications

Sample PCA, Tools for selecting the number of principal components, and Real data applications of PCA. Principal Component Analysis for Data Visualization

16

Lecture 16

L 16

Sparse PCA and Nonlinear Dimensionality Reduction

Sparse and robust methods for PCA, PCA for outlier detection, nonlinear dimensionality reduction, polynomial PCA, Basic elements of Nonparametric Density Estimation

17

Lecture 17

L 17

Kernel Principal Component Analysis

PCA for non-linear feature space, Kernel PCA,

18

Lecture 18

L 18

Latent Variable Model for Blind Source Separation

Latent variable models for blind source separation: cocktail party problem, independent component analysis (ICA) and its applications, linear mixing, and noiseless ICA

19

Lecture 19

L 19

ICA Algorithms and Exploratory Factory Analysis

FastICA algorithm for determining single source component, deflation, and parallel FastICA algorithm for extracting multiple independent source components, Applications to the real dataset, Exploratory factor analysis model

20

Lecture 20

L 20

Introduction to Artificial Neural Network

Basics and Structure of ANN, its various applications, ANN design and brain activity

21

Lecture 21

L 21

McCulloch- Pitts Neuron and Single-Layer Perceptron

Threshold logic unit, McCulloch-Pitts Neuron and its limitations, Hebb learning rule, Different types of neural networks

22

Lecture 22

L 22

Rosenblatt’s single-layer perceptron

Feedforward single layer network, Rosenblatt’s Single layer perceptron, single unit perceptron, Algorithm for implementing Rosenblatt’s single layer perceptron, perceptron convergence theorem

23

Lecture 23

L 23

Multi-layer perceptron

Multilayer perceptron, Learning networks, Multiclass classification rule

24

Lecture 24

L 24

Backpropagation of Errors Algorithm

Backpropagation of Errors Algorithm-Single hidden layer, Online learning mode, Stochastic learning mode, Batch learning mode

25

Lecture 25

L 25

Convolutional Neural Networks

Convolution Neural Network (CNN), its architecture, and applications

26

Lecture 26

L 26

Recurrent neural network and Projection Pursuit

Recurrent neural network (RNN) and CNN, Basics of RNN, Elman and Jordan networks, Projection pursuit regression, generalized additive model

27

Lecture 27

L 27

Cluster Analysis: An Introduction

Basic elements and objectives of cluster analysis, various similarity and distance measures,

28

Lecture 28

L 28

Hierarchical Clustering Techniques

Distance measures for quantitative variables, Hierarchical clustering method, Agglomerative Hierarchical Clustering, Single linkage Clustering, Complete Linkage Clustering, Average Linkage Clustering,

29

Lecture 29

L 29

Centroid and Non-hierarchical Clustering Methods

Centroid Linkage Clustering, Steps for implementation of Agglomerative Hierarchical Clustering, Ward’s Hierarchical Clustering method, Partitioning Clustering: K-means clustering

30

Lecture 30

L 30

Partition around medoids (PAM) Clustering Algorithm

PAM Clustering Algorithm, K-medoid and PAM clustering algorithm, Fuzzy analysis, Selecting number of clusters, Silhouette and average Silhouette Method

31

Lecture 31

L 31

Self-Organizing Map

Self-organizing maps (SOM) or Kohonen neural network, on-line and batch versions of SOM algorithm, distance weight version, U-matrix, Hierarchical SOM, Quality measures

32

Lecture 32

L 32

Clustering based upon Mixture Models

Density-Based Clustering Methods, Clustering based on Gaussian Mixture Models, Expectation-Maximization Clustering algorithm

33

Lecture 33

L 33

Recursive Partitioning: Decision Trees

Components of decision tree classification and basic terminology, Attribute Selection Measures: information gain and entropy, Gini index and node impurity function, choosing the best split, pruning algorithm for classification trees., Recursive partitioning to grow a tree, mating

34

Lecture 34

L 34

Training and Pruning Decision Trees

Overfitting and pruning the tree, cost complexity pruning measure, Choosing the best pruned tree, Cross validation for selecting best subtree

35

Lecture 35

L 35

Regression Trees

Regression trees: Background, Basic terminology, Recursive partitioning for regression data, terminal node value and splitting strategy, pruning the tree and best pruned subtree

36

Lecture 36

L 36

Committee Machine and Random Forests

Committee Machine: Bagging tree-based classifiers and regression tree predictors, Boosting, ADABOOST algorithm for binary classification. Random Forests algorithm for regression or classification

37

Lecture 37

L 37

Support Vector Machine for Linear Separable Cases

Support vector machine (SVM) with linear separable case, obtaining optimal separating hyperplane for linear separable case, Karuh, Kuhn, Tucker conditions, Multiclass SVM as a series of binary problems

38

Lecture 38

L 38

Support Vector Machine for Linearly Non-Separable Cases

SVM for nonlinearly separable datasets, nonlinear SVM, kernel trick for nonlinearly separable datasets, SVM for regression, e-insensitive loss function and its optimization

39

Lecture 39

L 39

Block Clustering

Basics of Block clustering, Hartigan’s block-clustering algorithm, Bi-clustering, two-way ANOVA model for bi-clustering

40

Lecture 40

L 40

Plaid Models for Block Clustering

Plaid models for bi-clustering with examples