Skip to main content

Time Series Analysis & Forecasting

In this course, we discuss state of the art techniques for time series analysis and forecasting.

About This Course

In this course, you will learn the essentials to perform time series analyses. We start with an introduction where we discuss the omnipresence of time series in data science and lay out the course objectives. Secondly, we discuss the specificities of time series data and models, including the key concept of stationarity. Next, we zoom in on a descriptive analysis of time series data. Tools such as the correlogram, filtering and smoothing will aid you in exploring data properties at the beginning of your time series analysis. The fourth part of the course then introduces indispensable tools for modeling single stationary time series, namely the flexible and popular family of AutoRegressive Moving Average (ARMA) models. This is followed by a section on forecasting and evaluating forecast performance. In part six of the course, we turn to the universe of non-stationary time series, providing a discussion on trends, unit roots and the need for unit root tests as an essential part of any time series analysis. Next, we turn from univariate time series models to multivariate time series models where we discuss how to jointly model the dynamics between several time series. We extensively discuss both stationary multivariate time series models such as Vector AutoRegressions as well as non-stationary time series models such as Vector Error Correction Models thereby equipping you with the necessary tools to perform cointegration analysis. Lastly, we provide some initial pointers on how to handle “Big” time series data sets, thereby reporting upon our own research contributions. The course concludes with an overview of all covered topics and provides an outlook on other interesting, more advanced time series topics.

The course combines methodological and technical insights with practical implementation details and code examples on how to use R for time series analysis and forecasting. We hereby focus on applications, how to interpret your results and how to avoid common pitfalls in empirical research.

The course features around 5 hours of video lectures , various multiple choice questions, and references to background literature. A certificate signed by the instructors is provided upon successful completion. See Big Data for Time Series Modeling to get a free teaser of the course contents.

We can also come and teach this course on-site in classroom format. If interested, please mail us at:


The enrollment fee for this course is EUR 100 (VAT excl.) per participant. Payments are securely handled by PayPal. If you are a company in the European Union, then we can apply VAT reverse charge. For this, please mail your VAT number to Part of our course revenue is used towards funding organizations involvement in protecting and cleaning our oceans. See our about page to learn more about our mission statement.

After enrollment, participants will get 1 year unlimited access to all course material (videos, quizzes and certificate).


Before subscribing to this course, you should have a basic understanding of descriptive statistics (e.g., mean, median, standard deviation, histograms, scatter plots, etc.) and inference (e.g., confidence intervals, hypothesis testing). Previous R and Python experience is helpful but not necessary.

Course Outline

  • Introduction
    • Instructor
    • Course Outline
    • Course Overview
    • Disclaimer
  • Motivation
    • Introduction
    • Objectives
  • Time Series Data
    • Data Types
    • Time Series Data and Models
    • Stationarity
    • Time Series Data Patterns
    • Quiz
  • Descriptive Analysis of Time Series Data
    • Descriptive Statistics
    • Filtering and Smoothing
    • Quiz
  • Univariate Stationary Time Series Models
    • AutoRegressive Moving Average Models
    • ARMA Lag Selection via ACF and PACF
    • ARMA Lag Selection via Information Criter
    • ia
    • ARMA Models: An Example in R
    • ARMA Models: Final R
    • Remarks
    • Quiz
  • Forecasting and Forecast Evaluations
    • Forecasting
    • Forecasting with ARMA Model
    • Forecast Evaluation
    • Splitting the data
    • Metrics for Forecast Evaluations
    • Quiz
  • Non-stationarity: Trend and Unit Roots
    • What Are Trends?
    • Dangers of Ignoring Trends
    • The Random Walk
    • The Drunkard’s Walk
    • Stochastic Trends
    • Stochastic and/or Deterministic Trends?
    • Autoregressive Models: The Effect of Shocks
    • Dealing with Trends
    • Unit Root Testing
    • Advanced Unit Root Testing
    • Quiz
  • Multivariate Time Series Models
    • Multiple Time Series
    • Single-Equation Models: Static versus Dynamic
    • Single-Equation Modelling in R
    • Shortcomings of the ARDL Model
    • Vector AutoRegressive Models
    • VAR Models: Remark
    • VAR Models: An Example in R
    • VAR Models: Choosing the lag order
    • VAR Models: Forecasting
    • VAR Models: Granger Causality
    • VAR Models: Impulse Response Analysis
    • Multiple Non-Stationarity Time Series
    • Cointegration
    • Cointegration vs. Spurious Regression
    • Cointegration: Testing
    • Error-Correction Model
    • Vector Error-Correction Model
    • VECM: Estimation, Testing, Forecasting
    • VECM Modelling in R: Metal Prices
    • Recap: Overview of Multivariate Models
    • Quiz
  • Big Time Series Analysis
    • “High-dimensional” Data and Models
    • Low versus High-dimensionality
    • Least Squares versus Penalized Regressions
    • Lasso
    • Lasso for VAR models
    • High-dimensional Time Series with Unit Roots
    • Quiz
  • Wrapping up and Looking ahead

Course Staff

Nalan Basturk

Nalan Basturk

Nalan was born on 25 December 1982 in Kocaeli (Turkey) and she lives in The Netherlands since 2005. She likes traveling, playing the oud and biking. Ironically, she developed a passion for mountain biking in The Netherlands instead of her hometown that is surrounded by hills and mountains. Nalan is an associate professor at the Department of Quantitative Economics at Maastricht University, The Netherlands. Her research focuses on time series models and Bayesian inference. She is interested in applications in economics and finance as well as in other disciplines. Recently, she obtained an NWO Vidi funding for developing Bayesian inference methods for describing and explaining time series patterns in mental health data and she has been a co-applicant in a successful ZON-MW consortium grant for developing new statistical methods for mental health data.

Stephan Smeekes

Stephan Smeekes

Stephan was born in Maastricht (the Netherlands) on January 5, 1983. Stephan is a life-long fan of his – not very successful – local football team MVV Maastricht, where he is also part of the members council. He also follows road cycling and Formula 1 racing closely, having regularly attended races at Spa-Francorchamps before Max Verstappen caused huge traffic jams of Dutch fans.

Stephan is also associate professor at the Department of Quantitative Economics at Maastricht University. His research interests lie in the statistical analysis of time series data, focusing on trends, uncertainty quantification using the bootstrap and the analysis of large time series datasets, with applications in macroeconomics and climatology. Stephan currently has an NWO Vidi grant on developing inference methods for large time series and is an elected member of the Dutch national Young Academy. He also appeared on Dutch national TV show “Knappe Koppen” to explain why predicting the next economic crisis is so difficult.

Etienne Wijler

Etienne Wijler

Etienne was born in Maastricht, The Netherlands, on March 20, 1991. He loves to ride his bike through the hilly landscapes of Limburg and sometimes pretends to be a professional by riding on an overqualified road bike and in uncomfortably tight clothes. In addition, Etienne loves cooking and eating, with exceptionally spicy Thai food being his favorite. He claims to choose a good Belgian beer over an expensive wine anytime, except when pasta is on the table.

Etienne Wijler is also a post-doctoral researcher at the Department of Quantitative Economics at Maastricht University. His research focuses on the development and statistical analysis of time series models for complex datasets. He particularly enjoys incorporating modern sources of “Big Data” into his applications, such as Google Trends data for explaining current unemployment figures or satellite data for the prediction of NO2 concentrations. Etienne has taught and (co-)developed numerous courses on data analytics and enjoys the interactions with students of different backgrounds. While his preferred research is of a purely theoretical nature, he aims to always translate his research into something that is easily interpretable and implementable, for example via the development of complementary R packages such as “specs”.

Ines Wilms

Ines Wilms

Ines was born in Anderlecht on August 6, 1989. Despite her birthplace she’s not-at-all a fan of the corresponding football team but she does like to cheer for the Red Devils on big tournaments. When it comes to practicing sports herself, she can easily be convinced for a ski trip (preferably to Austria)! Besides, she’s a big fan of desserts, mainly of consuming them but she also enjoys making them.

Ines Wilms is also assistant professor at the Department of Quantitative Economics of Maastricht University (the Netherlands). Her research focuses on developing statistical learning methods for analyzing large and complex data sets. She is particularly interested in analyzing large time series data sets, a topic for which she recently obtained funding via the H2020 Marie Skłodowska-Curie action BigTime. Demonstrating the relevance and usefulness of her developed methods for a wide variety of application domains including marketing, macro-economics and finance is an important part of her research. Ines has taught time series courses to students with diverse backgrounds at KU Leuven, Cornell University and Maastricht University.