Authors:
Kamil Dedecius
1
and
Pavel Ettler
2
Affiliations:
1
Institute of Information Theory and Automation and Academy of Sciences of the Czech Republic, Czech Republic
;
2
COMPUREG Plzen and s.r.o., Czech Republic
Keyword(s):
Statistical Analysis, Bayesian Analysis, Truncated Distributions, Beta Distribution.
Related
Ontology
Subjects/Areas/Topics:
Informatics in Control, Automation and Robotics
;
Information-Based Models for Control
;
Modeling, Analysis and Control of Discrete-event Systems
;
Signal Processing, Sensors, Systems Modeling and Control
;
System Modeling
Abstract:
Statistical analysis and modelling of various phenomena are well established in nowadays industrial practice.
However, the traditional approaches neglecting the true properties of the phenomena still dominate. Among
others, this includes also the cases when a variable with bounded range is analyzed using probabilistic distributions
with unbounded domain. Since many of those variables nearly fulfill the basic conditions imposed
by the chosen distribution, the properties of used statistical models are violated rather rarely. Still, there are
numerous cases, when inference with distributions with unbounded domain may lead to absurd conclusions.
This paper addresses this issue from the Bayesian viewpoint. It briefly discusses suitable distributions and
inferential methods overcoming the emerging computational issues.