ARTIFICIAL INTELLIGENCE

AND MACHINE LEARNING

APPROACH TO ARTIFICIAL INTELLIGENCE & MACHINE LEARNING
Udyog Bharat is making significant strides in introducing AI and ML
curricula to bolster India's technological and industrial capabilities. The
initiative aims to provide comprehensive education and training in Artificial
Intelligence (AI) and Machine Learning (ML) through various academic
programs and collaborations with leading educational institutions.

OBJECTIVES
Ÿ Acquire advanced Data Analysis skills.
Ÿ Stay Industry relevant and grow in your career.
Ÿ Create AI/ML solutions for various business problems.
Ÿ Build and deploy production grade AI/ML applications.
Ÿ Apply AI/ML methods, techniques and tools

COVERAGE

Business Case Studies Foundations of AI/ML

Data Visualization
10% 7%
5% Data Management
8%
20%
Artificial Intelligence 10% Statistical Thinking

20% 20%

Machine Learning Predictive Analytics

AI a multidisciplinary field that requires a range of skills in statistics, mathematics, predictive modeling and
business analysis. An AI professional should feel at ease to build the algorithms necessary, work with various
data sources (often in disparate forms) and an innate ability to ask the right questions and find the right answer.
This module helps layout the canvas on which the rest of the modules are built.

Unit 1: Introduction to Data Science and AI & ML

Ÿ Data Science, AI & ML
Ÿ Use Cases in Business and Scope
Ÿ Scientific Method
Ÿ Modeling Concepts
Ÿ CRISP-DM Method

Unit 2: R Essentials (Tutorial)

Programming
Ÿ Commands and Syntax
Ÿ Packages and Libraries
Ÿ Introduction to Data Types
Ÿ Data Structures in R - Vectors, Matrices, Arrays, Lists, Factors, Data Frames
Ÿ Importing and Exporting Data.
Ÿ Control structures and Functions

Descriptive Statistics
Ÿ Data exploration (histograms, bar chart, box plot, line graph, scatter plot)
Ÿ Qualitative and Quantitative Data
Ÿ Measure of Central Tendency (Mean, Median and Mode),
Ÿ Measure of Positions (Quartiles, Deciles, Percentiles and Quantiles),
Ÿ Measure of Dispersion (Range, Median, Absolute deviation about median, Variance and
Standard deviation), Anscombe's quartet
Ÿ Other Measures: Quartile and Percentile, Interquartile Range

Unit 3: Statistical Analysis

Initial Data Analysis
Ÿ Relationship between attributes: Covariance, Correlation Coefficient, Chi Square
Ÿ Measure of Distribution (Skewness and Kurtosis), Box and Whisker Plot (Box Plot and its parts,
Using Box Plots to compare distribution) and other statistical graphs

Probability
Ÿ Probability (Joint, marginal and conditional probabilities)
Ÿ Probability distributions (Continuous and Discrete)
Ÿ Density Functions and Cumulative functions
This is foundational to Data Scientists. This requires a nontrivial understanding of the real-world problems. It
involves judgments such as those about the relevance and representativeness of the data. This module helps
participants to have a good understanding of the methods, methodologies and techniques from the basics of
statistics and probability obtain supporting evidence through data, isolate or identify factors to construct
models that can uncover relationships and variation in processes.

Unit 3: Data Acquisition

Ÿ Gather information from different sources.
Ÿ Internal systems and External systems.
Ÿ Web APIs, Open Data Sources, Data APIs, Web Scrapping
Ÿ Relational Database access (queries) to process/access data.

Unit 4: Data Pre-processing and Preparation

Ÿ Data Munging, Wrangling
Ÿ Plyr packages
Ÿ Cast/Melt

Unit 5: Data Quality and Transformation

Ÿ Data imputation
Ÿ Data Transformation (minmax, log transform, z-score transform etc.,).
Ÿ Binning, Classing and Standardization.
Ÿ Outlier/Noise& Anomalies

Unit 6: Handling Text Data

Ÿ Bag-of-words
Ÿ Regular Expressions
Ÿ Sentence Splitting and Tokenization
Ÿ Punctuations and Stop words, Incorrect spellings
Ÿ Properties of words and Word cloud
Ÿ Lemmatization and Term-Document TxD computation
Ÿ Sentiment Analysis (Case Study)

Unit 7: Principles of Big Data

Ÿ Introduction to Big Data
Ÿ Challenges of processing Big Data (Volume, Velocity and Variety perspective)
Ÿ Use Cases

Unit 8: Big Data Frameworks – Hadoop, Spark and NoSQL

Ÿ Processing, Storage andProgramming Framework
Ÿ Hadoop eco-system Components and their functions
Ÿ Essential Algorithms (Word count, Page Rank, IT-IDF)
Ÿ Spark: RDDs, Streaming and Spark ML
Ÿ NoSQL concepts (CAP, ACID, NoSQL types)
This is foundational to Data Scientists. This requires a nontrivial understanding of the real-world problems. It
involves judgments such as those about the relevance and representativeness of the data. This module helps
participants to have a good understanding of the methods, methodologies and techniques from the basics of
statistics and probability obtain supporting evidence through data, isolate or identify factors to construct
models that can uncover relationships and variation in processes.

Unit 9: Data Visualization

Visualizing and Communicating clearly and effectively
about the patterns we find in data is a key skill for a
successful data professional. This module focuses on
the design and implementation of complementary
visual and verbal representations of patterns and
analyses in order to convey findings, answer questions,
drive decisions, and provide persuasive evidence
supported by data.

Ÿ Science of Visualization
Ÿ Visualization Periodic Table
Ÿ Aesthetics and Story telling
Ÿ Concepts of measurement - scales of measurement
Ÿ Design of data collection formats with illustration
Ÿ Principles of data visualization - different methods
of presenting data in business analytics.
Ÿ Concepts of Size, Shape, Color
Unit 11: Inferential Statistics
Ÿ Various Visualization types
Ÿ Develop an intuition how to understand the data,
Ÿ Bubble charts
attributes, distributions
Ÿ Geo-maps (Chlorpeths)
Ÿ Procedure for statistical testing, etc.
Ÿ Gauge charts
Ÿ Test of Hypothesis (Concept of Hypothesis
Ÿ Tree map
testing, Null Hypothesis and Alternative
Ÿ Heat map
Hypothesis)
Ÿ Motion charts
Ÿ Cross Tabulations (Contingency table and their
Ÿ Force Directed Charts etc.,
use, Chi-Square test, Fisher's exact test),
Ÿ One Sample t test (Concept, Assumptions,
Unit 10: Sampling and Estimation Hypothesis, Verification of assumptions,
Ÿ Sample versus population Performing the test and interpretation of results)
Ÿ Sample techniques (simple, stratified, clustered, Ÿ Independent Samples t test
random) Ÿ Paired Samples t test
Ÿ Sampling Distributions
Ÿ One way ANOVA (Post hoc tests: Fisher's LSD,
Ÿ Parameter Estimation
Tukey's HSD).
Ÿ Unbalanced data treatment
Ÿ z-test and F-test
Predictive analytics is an area of statistics that deals with extracting information from data and using it to
predict trends and behavior patterns. Predicting an outcome, predicting counts, predicting a value - all these
have immumerable use cases in CRM, rFaud detection, Portfolio Management, Sales and Marketing. Predictic
Analytics is approached from Regression (glm) and Time Series models in this module.

Unit 12: Linear Regression

Ÿ Regression basics: Relationship between attributes using Co variance and Correlation
Ÿ Relationship between multiple variables: Regression (Linear, Multivariate) in prediction.
Ÿ Residual Analysis
Ÿ Identifying significant features, feature reduction using AIC, multi-collinearity
Ÿ Non-normality and Heteroscedasticity
Ÿ Hypothesis testing of Regression Model
Ÿ Confidence intervals of Slope
Ÿ R-square and goodness of fit
Ÿ Influential Observations – Leverage

Unit 13: Multiple Linear Regression

Ÿ Polynomial Regression
Ÿ Regularization methods
Ÿ Lasso, Ridge and Elastic nets
Ÿ Categorical Variables in Regression

Unit 14: Non-Linear Regression

Ÿ Logit function and interpretation
Ÿ Types of error measures (ROCR)
Ÿ Logistic Regression in classification

Unit 15: Forecasting models

Ÿ Trend analysis
Ÿ Cyclical and Seasonal analysis
Ÿ Smoothing; Moving averages; Box-Jenkins, Holt-winters, Auto-correlation; ARIMA
Ÿ Examples: Applications of Time Series in financial markets
Predictive analytics is an area of statistics that deals with extracting information from data and using it to
predict trends and behavior patterns. Predicting an outcome, predicting counts, predicting a value - all these
have immumerable use cases in CRM, Fraud detection, Portfolio Management, Sales and Marketing. Predictic
Analytics is approached from Regression (glm) and Time Series models in this module.
Unit 16: Foundations for ML
Ÿ ML Techniques overview
Ÿ Validation Techniques (Cross-Validations)
Ÿ Feature Reduction/Dimensionality reduction
Ÿ Principal components analysis (Eigen values, Eigen vectors, Orthogonality)

Unit 17: Clustering

Ÿ Distance measures
Ÿ Different clustering methods (Distance, Density, Hierarchical)
Ÿ Iterative distance-based clustering;
Ÿ Dealing with continuous, categorical values in K-Means
Ÿ Constructing a hierarchical cluster
Ÿ K-Medoids, k-Mode and density-based clustering
Ÿ Measures of quality of clustering
Unit 18: Classification
Naïve Bayes Classifier
Ÿ Model Assumptions, Probability estimation
Ÿ Required data processing
Ÿ M-estimates, Feature selection: Mutual information
Ÿ Classifier

K-Nearest Neighbors
Ÿ Computational geometry; Voronoi Diagrams; Delaunay Triangulations
Ÿ K-Nearest Neighbor algorithm; Wilson editing and triangulations
Ÿ Aspects to consider while designing K-Nearest Neighbor

Support Vector Machines

Ÿ Linear learning machines and Kernel space, Making Kernels and working in feature space
Ÿ SVM for classification and regression problems.

Decision Trees
Ÿ ID4, C4.5, CART

Ensembles methods
Ÿ Bagging & boosting and its impact on bias and variance
Ÿ C5.0 boosting
Ÿ Random forest
Ÿ Gradient Boosting Machines and XGBoost

Unit 19: Association Rule mining

Ÿ The applications of Association Rule Mining: Market Basket, Recommendation Engines, etc.
Ÿ A mathematical model for association analysis; Large item sets; Association Rules
Ÿ Apriori: Constructs large item sets with mini sup by iterations; Interestingness of discovered association
rules;
Ÿ Application examples; Association analysis vs. classification
Ÿ FP-trees
Artificial Intelligence is utilized heavily in computizing cognitive functions such as speech and Vision. Often
these functions are achieved through the use of Neural networks. In this module, we will study very popular NN
architectures for achieving various cognitive functions such as Object recognition, natural language processing
besides explore reinforcement learning. We will study and practice various use cases ranging from text
generation, object tagging to fraud detection and learning games such as 2048.

Unit 20: Foundations for AI

Ÿ AI: Application areas
Ÿ AI Basics (Divide and Conquer
, Greedy, Branch and Bound, Gradient Descent)
Ÿ NN basics (Perceptron and MLP, FFN, Backpropagation)

Unit 21: Convolution Neural Networks [2]

Ÿ Image classification
Ÿ Text classification
Ÿ Image classification and hyper-parameter tuning
Ÿ Emerging NN architectures

Unit 22: Recurrent Neural Networks [2]

Ÿ Building recurrent NN
Ÿ Long Short-Term Memory
Ÿ Time Series Forecasting

Unit 23: Deep Learning [6]

Ÿ Auto-encoders and unsupervised learning
Ÿ Stacked auto-encoders and semi-supervised learning
Ÿ Regularization - Dropout and Batch normalization
Case Study 1: Churn Analysis and Prediction (Survival Modelling)
Ÿ Cox-proportional models
Ÿ Churn Prediction

Case Study 2: Credit card Fraud Analysis

Ÿ Imbalanced Data
Ÿ Neural Network

Case study 3: Sentiment Analysis or Topic Mining from New York Times
Ÿ Similarity measures (Cosine Similarity, Chi-Square, N Grams)
Ÿ Part-of-Speech Tagging
Ÿ Stemming and Chunking

Case Study 4: Sales Funnel Analysis

Ÿ A/B testing
Ÿ Campaign effectiveness, Web page layout effectiveness
Ÿ Scoring and Ranking

Case Study 5: Recommendation Systems and Collaborative filtering

Ÿ User based
Ÿ Item Based
Ÿ Singular value decomposition–based recommenders

Case Study 6: Customer Segmentation and Value

Ÿ Segmentation Strategies
Ÿ Lifetime Value

Case Study 7: Portfolio Risk Conformance

Ÿ Risk Profiling
Ÿ Portfolio Optimization

Case Study 8: Uber Alternative Routing

Ÿ Graph Construction
Ÿ Route Optimization