Introduction to AI

Quick Info: Bachelor | Lecture | English | Winter Semester | 6 ECTS

Summary

Course	Introduction to AI
Chair	Institute of AI in Management
Lecturer	Prof. Dr. Stefan Feuerriegel
Assistants	Annually changing
Weekly Hours	This is a block course with approximately five full days of lectures and exercises, plus two half days for Q&A, during the semester.
Target Group	B.Sc. BWL
Examination	We plan for an exam on site, in person. The exam will include programming exercises, to large extent.
Prerequisites	We expect scripting skills (e.g., R or Python), as students will apply their theoretical knowledge by implementing a machine learning application with given open-source packages. Please bring your own laptop to class.
Course Material	Course material will be shared via Moodle, students are required to self-enrol to the course through Moodle, self-enrolment key can be accessed via LSF.
Rotation	Annual rotation, winter semester
Language	English
ECTS	6

LSF
Moodle

Course Overview

Description

Prior to the start of the Information Age in the late 20th century, companies were forced to collect data from non-automated sources manually. Companies back then lacked the computing capabilities necessary for data to be analysed, and as a result, decisions pri-marily originated not from knowledge but from intuition.

By the end of the course, students will be able to plan, implement and evaluate analytics in applied settings to generate value from data for society, corporations, and individuals. This serves the pressing need of firms to improve their efficiency – such as customer satisfaction, competitive advantage – by leveraging the growing amounts of structured and unstructured data.

To achieve this overall goal, students should after participation being able to:

Objective 1 (Managerial aspects): Understand the processes and challenges of analytics-related projects

Identify applications for analytics in corporations and organizations that create value
List implications for management when undertaking a project involving business analytics
Apply the data mining process CRISP-DM to their actual setting

Objective 2 (Methodological challenges): Understand common methods for performing business analytics

Translate use cases of business analytics into a mathematical model formulation
Name common methods for business analytics, as well as their underlying concepts
Compare the properties of these models

Objective 3 (Practical implementation): Performing actual evaluations of business analytics based on real-word datasets

Pre-process data to transform it into relational structures

Apply statistical software (e.g., R or Python) to perform business analytics in practice

Evaluate the results to choose the best-performing method

Outline

Final dates and times will be shared via Moodle.

Unit 1: Introduction

Organizational details
Applications of business analytics

Unit 2: Predictive modeling

Definition of machine learning
Taxonomy of predictive modeling
Performance assessments

Unit 3: Linear modeling

Linear model (ordinary least squares)
Regularization (lasso, ridge regression, elastic net)

Unit 4: Non-linear modeling

Decision trees
Random forest
Boosting
Neural networks

Unit 5: Model tuning

Train/test split
Cross-validation

Unit 6: Bringing machine learning to practice

Management challenges
Pitfalls in practice

Beyond the scope of the course: resampling, unsupervised learning (clustering), dimen-sionality reduction (principal component analysis), classifiers (support vector machine, AdaBoost, etc.)

Literature

James, Witten, Hastie & Tibshirani (2013): An Introduction to Statistical Learning: With Applications in R. Springer
Sharda, Delen & Turban (2014): Business Intelligence: A Managerial Perspective on Ana-lytics. Pearson