Applied Forecasting in Complex Systems

Course manual 2022/2023

Course content

From the beginning, students will learn how to analyze and forecast the dynamics of large-scale real-complex time-series data, from social or technological systems and forecast. Students will learn how to describe time-series prediction methods and the applications of these methods to different types of data in various contexts. Risk and uncertainty are central to forecasting and prediction; it is generally considered good practice to indicate the degree of uncertainty attaching to forecasts, and often it is necessary to provide distributional rather than point forecasts.  As such, an introduction to methods for probabilistic forecasting will also be provided. Time series modeling techniques will be considered with reference to their use in forecasting where suitable. While linear models will be examined in some detail, extensions to non-linear models using machine learning approaches will also be considered for real-world problems. In the course, students will learn the theoretical concepts during the lectures as exploring different problem domains, build up their skills by practicing on assignments, and finally demonstrate their knowledge and skills by a final project.

Study materials

Literature

• Hyndman, R.J., & Athanasopoulos, G. (2018) Forecasting: principles and practice, 2nd edition, OTexts: Melbourne, Australia.

• Principles of Business Forecasting, 2nd ed. by Keith Ord, Robert Fildes, Nikolaos Kourentzes

• Practical Time Series Analysis: Prediction with Statistics and Machine Learning, 1st Edición

Syllabus

• A syllabus containing articles and chapters will be made available at the beginning of the course. The syllabus/course will cover parts of the books proposed in the literature.

Software

• R and Python

Objectives

• Explain to what extent results derived from forecasts are useful to the problem at stake
• Review and interpret modeling and forecasting results critically
• Discuss the potential setup of forecasting models
• Analyze the visualization of uncertainty attaching to forecasts, and how it relates and differentiates from point forecasts
• Select the most appropriate forecasting method and devise time series prediction models suitable for solving real-problem

Teaching methods

• Lecture
• Computer lab session/practical training
• Self-study
• Seminar

Learning activities

Attendance

In TER part B of this programme no requirements regarding attendance are mentioned.

Additional requirements for this course:

Absence needs to be communicated to the course coordinator when not doing the course.

Assessment

• The course grade G is based on the quiz grades Q1, Q2, Q3, assignment grades A1, A2,  and the Final Project grade FP.

• If Assignment i is not submitted, Ai=0.

• 3 online quizzes (20%), 1 assignment (individual) (20%), 1 assignment (group) (20%), Final Project (40%).  
•  To pass, the grades have to be 5.50 or higher in both the individual and the group cluster (i.e., minimum 5.50 on the average of the FP and the group assignment and minimum 5.50 on the average of both individual assignment and the quizzes).

Assignments

There will be 2 graded assignments. One is individual the other is a group (~4 people).  In addition, there is a final group project where a written report should be delivered.

Fraud and plagiarism

The 'Regulations governing fraud and plagiarism for UvA students' applies to this course. This will be monitored carefully. Upon suspicion of fraud or plagiarism the Examinations Board of the programme will be informed. For the 'Regulations governing fraud and plagiarism for UvA students' see: www.student.uva.nl

Course structure

Timetable

The schedule for this course is published on DataNose.

Contact information

Coordinator

• dr. rer. nat. Erman Acar