GRA 8259 Machine Learning
GRA 8259 Machine Learning
Digitalization creates high speed, high volume digital activity traces – Big Data. Using these data and recent advances in machine learning (ML), organizations can automatically build predictive models and tools that can improve a wide variety of routines, ranging from targeted marketing to preventative maintenance to fraud detection. But adopting these tools requires: specifying viable problems, recognizing underlying data requirements and shortcomings, understanding models’ potential failure points, assessing model performance, translating model predictions to actionable business insights, and reflecting on the ethical and security implications of data collection/model building. To successfully deploy ML/AI, managers must become familiar and comfortable with foundational tasks. Managers do not need to become data scientists, but managers in digital organizations need to understand what ML can and cannot do, what ML requires, how to formulate viable ML questions, how to action data-driven predictions, and how models can break.
This course develops these foundational skills through hands-on experience with cutting edge analytics tools and techniques, and through case discussions designed to reflect on the technical foundations of ML/AI successes and failures.
- Candidates will understand the relationship between data structures, modeling, and problem specification.
- Using ML tools, candidates will understand potential modeling and data errors that can make models look good in development but fail in practice.
- Candidates will understand security issues and practices associated with machine learning and the use of machine learning for security.
- Candidates will be able to identify and be familiar with tools and ingredients required to start a machine learning project.
- Candidates will be able to identify viable machine learning problems.
- Candidates will be able to productively work with data science teams, specifying reasonable projects and assessing modeling outcomes.
- Candidates will examine social and security tensions inherent to modeling.
Course outline (on an overall level)
- Machine learning’s location within the analytics landscape
- Data structures, and the correspondence between data and ML tasks
- Specifying viable and valuable ML problems
- Modeling processes and techniques
- Automated ML
- ML case studies, reflecting on the technical underpinnings
The course consists of lectures, ML labs using Python (on local computers and cloud), and some case discussions. Lectures will reference Python code for illustrative purposes, but students are not expected to have any coding background. Experience demonstrates that reading code, and walking through its execution incrementally, is an accessible way to illustrate how ML modeling really works. For this, students will need to install Continuum’s Anaconda suite on their local machines. Transitioning from Python labs, the class will advance to cloud labs exploring automated machine learning applications. Students may need to purchase licenses depending on what type of cloud services we use for the labs. This will be informed well ahead. Following introductory lectures and labs, we will use applications and cases to reflect on how tools are used, applications for which tools might be developed, and ethical/security considerations.
Attendance to all sessions in the three core topics of the course is compulsory. If you have to miss part(s) of any of the three core topics of the course you must ask in advance for leave of absence. More than 25% absence in one of the core topics of the course will require retaking the entire topic. It's the student's own responsibility to obtain any information provided in class that is not included on the course homepage/ It's learning or other course materials.
Candidates may be called in for an oral hearing as a verification/control of written assignments.
All the three core topics of the course must be passed in order to obtain a grade for the course.
The course is a part of a full Executive MBA programme and examination in all courses must be passed in order to obtain a certificate.
| Assessments |
|---|
Exam category: Submission Form of assessment: Written submission Weight: 100 Grouping: Individual Duration: 4 Week(s) Comment: Written Assignment, counts 100% of the final grade. Exam code: GRA 82591 Grading scale: ECTS Resit: Examination when next scheduled course |
| Activity | Duration | Comment |
|---|---|---|
Teaching | 40 Hour(s) | |
Prepare for teaching | 35 Hour(s) | |
Student's own work with learning resources | 75 Hour(s) | Self study, feedback activities/counselling and exam |
A course of 1 ECTS credit corresponds to a workload of 26-30 hours. Therefore a course of 5 ECTS credit corresponds to a workload of at least 135 hours.