|
|
Maximum likelihood estimators in time series
Tritová, Hana ; Pawlas, Zbyněk (advisor) ; Zikmundová, Markéta (referee)
The thesis deals with maximum likelihood estimators in time series. The reader becomes familiar with three important models for time series: autoregressive model (AR), moving average model (MA) and autoregressive moving average (ARMA). Thereafter he can find out the form of their main characteristics, e.g. population mean and variance. Then there is the derivation of parameter estimates - generally and for mentioned models of times series. There are also stated two other methods for finding estimators of AR(1) and MA(1) parameters - method of moments and least squares method. The end is dedicated to examples which compares all three methods.
|
|
|
Statistical applications of urn models
Navrátil, Radim ; Pawlas, Zbyněk (advisor) ; Omelka, Marek (referee)
This work shows various applications of urn models in practice. First, basic properties of the occupancy distribution are derived together with its asymptotic approximation. This model is applied and generalized in the theory of database systems for records search from a given database. An application to random texts is mentioned, namely the computation of the expected number of missing and common words in random texts. There are presented exact formulas, their asymptotic approximations and the approximations via occupancy distribution. Then, some urn models, which are used in the randomized response theory for finding out respondents' answers to sensitive questions, are described. These models are compared according to their accuracy and respondents' goodwill to answer. Finally, two non-parametric tests of empty boxes are derived, one for the hypothesis whether a random sample comes from a given population and the second for the hypothesis whether two independent random samples come from the same population. The powers of these tests are compared with commonly used tests for these hypotheses.
|
|
|
Interacting spatial particle systems
Zikmundová, Markéta ; Beneš, Viktor (advisor) ; Pawlas, Zbyněk (referee) ; Volf, Petr (referee)
1 Title: Interacting spatial particle systems Author: Markéta Zikmundová Department: Department of Probability and Mathematical Statistics Author's e-mail address: zikmundm@karlin.mff.cuni.cz Supervisor: Prof. RNDr. Viktor Beneš, DrSc. Supervisor's e-mail address: benesv@karlin.mff.cuni.cz Consultant: RNDr. Kateřina Helisová, Ph.D. Consultant's e-mail address: helisova@math.feld.cvut.cz Abstract: Several kinds of random union of interacting particles is studied. We define line segment process of interacting particles in R2 and process of interacting surfaces in R3 as the models with density function p with respect to some Poisson point process. The formulas for moments of the geometrical characteristics of these models are derived and the limit behaviour when the intensity tends to infinity is investigated. For time extension of such models a simulation algorithm is developed. Various estimations of parameters of density p, among them those based on sequential Monte Carlo, are studied and compare in a simulation study. Keywords: Boolean model, process with interacting particles, U−statistics, exponential family, germ-grain model, interaction, Markov properties, point process, random closed set, Markov chain Monte Carlo.
|
|
| |
|
|
Sampling methods in forestry
Hanek, Petr ; Pawlas, Zbyněk (advisor) ; Jurečková, Jana (referee)
This diploma thesis is devoted to the sampling strategies in forestry. It describes their theoretical aspects and their applications on a real landscape. The sampling methods in forestry are of particular importance in forest inven- tory. The aim of sampling methods is to estimate population characteristics based on the knowledge of sample. Two basic approaches can be distinguished according to the size of population, we speak about discrete or continuous population. Several types of sampling designs and corresponding estimators of target values are described for both approaches. Besides estimates of po- pulation total or average, we mention the formulas for computing variance of these estimates and the methods for their estimation for different sampling designs. The thesis also contains the comparison of studied methods based on computer simulations.
|
|
| |
|
|
Spatial econometrics
Nývltová, Veronika ; Pawlas, Zbyněk (advisor) ; Kopa, Miloš (referee)
This thesis is devoted to the models that are suitable for modelling spatial data. For this purpose, random fields with finite index set are used. Based on the neighbourhood relationship a spatial weight matrix is introduced which describes spatial dependencies. A recognition and testing of spatial dependence is mentioned and it is applied for macroeconomic indicators in the Czech Republic. Spatial models originated from generalization of usual time series models are subsequently combined with linear regression models. The parameter estimators are derived for selected models by three different methods. These methods are ordinary least squares, maximum likelihood and method of moments. Theoretical asymptotic results are supplemented by a simulation study that examines the performance of estimators for finite sample size. Finally, a short illustration on real data is demonstrated. Powered by TCPDF (www.tcpdf.org)
|
|
| |
|
|
Bayesian modeling of market price using autoregression model
Šindelář, Jan ; Kárný, Miroslav (advisor) ; Pawlas, Zbyněk (referee) ; Šmíd, Martin (referee)
1 Bayesian modeling of market price using autoregression model 1Šindelář Jan Department: Department of Probability and Mathematical Statistics Supervisor: Ing. Miroslav Kárný, DrSc. Abstract: In the thesis we present a novel solution of Bayesian filtering in autoregression model with Laplace distributed innovations. Estimation of regression models with lep- tokurtically distributed innovations has been studied before in a Bayesian framework [2], [1]. Compared to previously conducted studies, the method described in this article leads to an exact solution for density specifying the posterior distribution of parameters. Such a solution was previously known only for a very limited class of innovation distributions. In the text an algorithm leading to an effective solution of the problem is also proposed. The algorithm is slower than the one for the classical setup, but due to increasing com- putational power and stronger support of parallel computing, it can be executed in a reasonable time for models, where the number of parameters isn't very high. Keywords: Bayesian, Autoregression, Optimal Trading, Time Series References [1] P. Congdon. Bayesian statistical modelling. Wiley, 2006. [2] A. Zellner. Bayesian and Non-Bayesian analysis of the regression model with multivari- ate Student-t error term. Journal...
|
|
|
Set-indexed stochastic processes
Schenk, Martin ; Pawlas, Zbyněk (advisor) ; Rataj, Jan (referee)
This thesis deals with the problem of estimating the joint probability distribution of a marked process' parameters from a censored data. First, a Nelson-Aalen estimator of the cumulative hazard rate for one-dimensional case is constructed. This estimator is then smoothed by using a kernel function estimator. Then, a Kaplan-Meier estimator of the survival function is brought in. Further, a theory of set-indexed random processes is built up to be a base for the construction of a generalized Nelson-Aalen estimator of the cumulative hazard rate, which is then again smoothed. For a special case, a generalized Kaplan-Meier estimator of the multidimensional survival function is constructed. The application of the mentioned generalized estimators is shown on a particular case. These estimators are then used on simulated data.
|
guest ::
