University of Twente Student Theses


An in-depth approach to fitting probability distributions using evolutionary algorithms

Constantinescu, E.M. (2020) An in-depth approach to fitting probability distributions using evolutionary algorithms.

[img] PDF
Abstract:The practice of estimating probability density functions is well-known in the field of statistics and probability theory, with commonly used techniques including histograms, kernel estimators, orthogonal series and nearest-neighbor methods. Furthermore, goodness-of-fit tests are being used for evaluating the accuracy of a statistical model, based on a provided set of observations. While these practices have been widely used so far, they heavily rely on human assumptions and deductions and. This causes them to be susceptible to errors and bias. Techniques using evolutionary algorithms - a metaheuristic optimization type of algorithms - have had optimistic results in estimating the parameters of an assumed distribution function and, as later works have shown, determining the most probable distribution type with its respective parameters. However, recent work scarcely provide any discussions about the importance of the design choices behind important aspects of the evolutionary algorithm. With these facts in mind, the goals of this paper are to explore the alternative of using an evolutionary algorithm to not only estimate the parameters of a probability distribution but to also choose which type of distribution shape would fit the observed data best and to provide an in-depth discussion which will allow for easier replicability.
Item Type:Essay (Bachelor)
Faculty:EEMCS: Electrical Engineering, Mathematics and Computer Science
Subject:31 mathematics, 54 computer science
Programme:Computer Science BSc (56964)
Link to this item:
Export this item as:BibTeX
HTML Citation
Reference Manager


Repository Staff Only: item control page