TILS681 Nonparametric and Robust Methods (5 cr)
Description
Classical statistical methods based on the mean are optimal only when the assumption of a normal distribution holds. The objective of this course is to introduce alternative methods that remain valid and efficient across a broader range of statistical models. To achieve this, we will delve into general concepts of non-parametric and robust statistical methods, with a primary focus on regression models.
Traditional non-parametric and robust linear regression models relax distributional assumptions while retaining the linearity assumption. To further relax the linearity assumption, we will also explore regression methods based on kernels and splines during the course.
The utilization of these methods and the exploration of their properties will be practiced using the R software.
Learning outcomes
Upon successfully completing the course, the student will:
- have knowledge of the fundamental theory behind nonparametric and robust methods,
- be able to investigate the properties of these methods through computer simulations,
- be capable of selecting an appropriate method for empirical data and applying it using the R software.
Additional information
The course will be lectured at most every two years.
Description of prerequisites
Statistical Inference 1 and 2, Generalized linear models 2, some basic knowledge of R