Course

Discipline Specific Electives: Business Analytics

Unit 1

Unit 1 – Introduction to Machine Learning

Introduction to Machine Learning, Broad classification – Supervised vs Unsupervised Learning, Use cases of Machine Learning,

Data cleansing: Treating missing values, Treating outliers and errors, Data virtualization – a unified view of business entities from multiple sources of data, Data lake – Ingestion of multiple sources of data for data transformation

Unit 2

Unit 2 – Regression

Introduction to Regression – Simple & Multiple Regression, Estimation, Goodness of fit measures, Diagnostics, Binary Logistic Regression, Model validation, Applications

Unit 3

Unit 3 – Classification and Validation Measures

Classification using Decision Trees, Random Forests, Classification Nearest Neighbours, Classification using Naïve Bayes, Goodness measures such as confusion matrix, Validation (Bagging and Boosting), and Applications.

ROC curves- comparison of distribution function/business measures, Divergence, Kolmogorov Smirnov- difference in distribution function, Gini coefficient/D concordance statistic.

Unit 4

Unit 4 – Clustering

Introduction to Clustering, K-means, Hierarchical Clustering, Practical Issues in Clustering, Validation, and Applications.

Unit 5

Unit 5 – Recommendation Systems and Customer Analytics

Collaborative filtering such as User-based, Item-based, and Matrix Factorization; Evaluation of Recommenders such as Cumulative gain and discounted cumulative gain. Applications of recommendation systems and association Rules such as the Apriori Algorithm.

Objective:

This course helps students understand the techniques and application of machine learning. Course Outcome

CO1: Apply the techniques of data transformation and data reduction.

CO2: Implement a simple linear regression model in Python. Assess the performance of a simple/multi-linear regression model.

CO3: Build and diagnose a supervised learning algorithm with quantitative response variables. CO4: Understand the workings of unsupervised learning algorithms.

CO5: Use relevant machine learning algorithms based on business data and problems.

Customer Life Cycle, Segmentation, Scoring, Use cases References Textbook:

• Data Mining for Business Analytics: Concepts, Techniques, and Applications with XLMiner, 3rd Edition