Decision Making under Complex Uncertainty

Marc Johler
The aim of this work is the presentation and implementation of a selection of principles and criteria of decision theory in the statistical programming language R. The selection contains the most common criteria as well as some criteria introduced by the chair for "Foundations of Statistics and Their Applications". Theoretical aspects and background information on the criteria is included, as well as explanations for each of the their implementations. Most of the algorithms used did already
more » ... t, still this thesis contains suggestions of new approaches to compute some of the criteria. The chosen algorithms will be critically evaluated and compared to possible alternatives in terms of running time. Moreover the functions and their input structure have been defined in a way which should enable the decision maker to insert her information about the acts, states and the used utility or loss function in a straight-forward and convenient way. Many approaches in this thesis are based on the idea of linear programming. The algorithms for solving these kind of problems are not part of this thesis, since one can rely on already existing linear-programming-packages like rcdd and lpSolve. For the manipulation of the data and stylistical aspects of the code the author made use of the functionality of the tidyverse package. The checks of input validity have been partially done with the package checkmate.
doi:10.5282/ubm/epub.75183 fatcat:af2jkuayl5ahdag43ybq4qfx4m