GRA 4136 Predictive Analytics and Machine Learning
GRA 4136 Predictive Analytics and Machine Learning
This course teaches students how to use machine learning techniques and tools for predictive analytics. The course will also teach students to: specify problems accessible with these methods, evaluate analytics models, use new tools in the rapidly evolving automated machine learning field, and deploy machine learning models. The course assumes prior experience with: basic probability theory, traditional methods for analysis of continuous and categorical variables, and introductory Python programming.
Students will learn about machine learning algorithms, translating business problems into problems machines can solve, and evaluating machine learning models. Students will also learn about recent advances in automated machine learning, and the implications for generalized deployment of the technology. Students will extend their programming expertise to machine learning applications, using predictive analytics to address business problems and to evaluate strategic options. Students will become familiar with automated machine learning tools such as: Python’s auto_ml and tpot libraries, DataRobot, exploratory.io, etc. Because these automated ML tools are very new and changing very rapidly, final decisions on tools will depend on the “state of the art” when the course starts.
Students completing the course will be able to: specify business problems such that predictive analytics can be applied; evaluate and specify project data requirements; execute predictive analytics on a broad range of problems; evaluate predictive analytics model results; and discuss results/requirements both with data scientists and with practitioners.
Positioning data vis-à-vis methods, as in predictive analytics, is a natural opportunity to inculcate a perspective that organizations are problem-solving, data-driven entities. Predictive analytics offers a framework in which students can reflect on the value of data for a particular problem, or the value of a method given particular data. The course will encourage students to reflect on capturing and creating data-driven value. Because analytics is data-driven, as compared to traditional theory-driven research, the course will ask students to reflect on the strengths and weaknesses of each scientific approach.
This course focuses on predictive analytics and machine learning applications: supervised learning (classification and regression). The course will also introduce unsupervised learning, including clustering, market basket analysis, and dimensionality reduction. Students will implement and tune individual machine learning algorithms (e.g., via Python’s scikit-learn library), and work with tools in the rapidly evolving field of automated machine learning. In addition to lecture, the course will include cases, in-class workshops, and guest speakers as appropriate.
The course will use a variety of software, including: Continuum's Anaconda software suite, DataRobot, and MySQL. This field is changing very rapidly and new tools are released monthly. To ensure that the course stays current, the instrutor may introduce new software. Such changes will not require any prerequisite changes or fundamentally alter the nature of the course.
During class, the instructor will provide students with a link to purchase a 9-month DataRobot license. At writing, this license costs $80, and is paid for by the student.
Please note that while attendance is not compulsory in all courses, it is the student’s own responsibility to obtain any information provided in class.
All parts of the assessment must be passed in order to get a grade in the course.
All courses in the Masters programme will assume that students have fulfilled the admission requirements for the programme. In addition, courses in second, third and/or fourth semester can have specific prerequisites and will assume that students have followed normal study progression. For double degree and exchange students, please note that equivalent courses are accepted.
The course assumes prior experience with: basic probability theory, traditional methods for analysis of continuous and categorical variables, and introductory Python programming.
| Assessments |
|---|
Exam category: Submission Form of assessment: Written submission Weight: 50 Grouping: Individual Duration: 72 Hour(s) Comment: Machine learning take-home exam. Exam code: GRA 41361 Grading scale: ECTS Resit: Examination when next scheduled course |
Exam category: Submission Form of assessment: Written submission Invigilation Weight: 50 Grouping: Individual Support materials:
Duration: 3 Hour(s) Comment: Written examination under supervision. Exam code: GRA 41362 Grading scale: ECTS Resit: Examination when next scheduled course |
All exams must be passed to get a grade in this course.
A course of 1 ECTS credit corresponds to a workload of 26-30 hours. Therefore a course of 6 ECTS credits corresponds to a workload of at least 160 hours.