Analyzing time series data cannot be done by ordinary least squares regression techniques due to serial correlation. To model such dependencies in an accessible way state space models provide a flexible framework that includes many classical approaches such as AR(I)MA or structural time series models.
In this workshop you will learn
- to model time series data with state space models,
- to implement these models in the statistics software R with the KFAS package (https://github.com/helske/KFAS),
- to interpret the results of the fitted models and
- to quantify uncertainty in these models.
The datasets used in this workshop will largely stem from computational communication science.
Some familiarity with R is required to participate in this workshop. For example, the contents of the workshops "Statistics with R" or "Introduction to R" are sufficient to participate. Participation is free of charge for members of the EI, IA, MN and WM departments. Please note that members of the MB department cannot participate in the workshops of the statistical consulting services.
- Teacher: Stefan Heyder