|
|
Tests in multinomial distribution
Holý, Vladimír ; Anděl, Jiří (advisor) ; Antoch, Jaromír (referee)
In this paper there are at first described classical goodness-of-fit tests - the Pear- son's χ2 test and the log likehood ratio test. The more modern method of testing is the family of statistics based on power divergence which is generalisation of classical statistics. Another type of generalisation is the family of disparity statis- tics which includes beside the family of power divergence also the families BWHD and BWCS. It is demonstrated that all these test statistics have an asymptotic χ2 distribution. In the program R the exact level and exact power can be calculated for individual tests. Hereafter, moments of test statistics can be derived. On the basis of these comparisons there will be shown which test statistics are the most suitable for the goodness-of-fit tests. 1
|
|
|
Confidence Intervals for Quantiles
Horejšová, Markéta ; Kulich, Michal (advisor) ; Hlávka, Zdeněk (referee)
In this thesis, various construction methods for simultaneous confidence intervals for quantiles are explained. Among nonparametric approaches, a special emphasis is dedicated to a recent method based on a multinomial distribution for calculating the overall confidence level of confidence intervals for all quantiles of interest using an efficient recursive algorithm, which is also described. Furthermore, a method based on Kolmogorov-Smirnov statistic or an asymptotic method using empirical distribution function and order statistics for quantile estimate are presented. A special parametric method for several quantiles of a normally distributed population is introduced along with a few of its modifications. Subsequently, a simulation is run to test the real coverage of the described theoretical methods. Powered by TCPDF (www.tcpdf.org)
|
|
|
Confidence intervals for parameters of multinomial distribution
Bárnetová, Kamila ; Anděl, Jiří (advisor) ; Omelka, Marek (referee)
Title: Confidence intervals for parameters of multinomial distribution Author: Kamila Bárnetová Department: Department of Probability and Mathematical Statistics Supervisor: prof. RNDr. Jiří Anděl, DrSc., Department of Probability and Mathematical Statistics Abstract: Confidence intervals for parameters for binomial and multinomial distribution are described in this thesis. These intervals can be used in practice, for exemple- pre-election estimates. The first two chapter are devoted to derivation of these intervals. Simulations and comparison of several selected methods can be found in the last chapter. Based on the simulations, we consider it appropriate, to choose to calculate confidence intervals for parameters of multinomial distribution intervals based on Bonferroniho inequality, or their modifications. These intervals are easy to calculate, while their coverage probability is at least 0.89. Keywords: confidence interval, multinomial distribution, binomial distribution, Bonferroni inequality
|
|
|
Distribution of interpoint distances
Horská, Šárka ; Hlávka, Zdeněk (advisor) ; Komárek, Arnošt (referee)
This thesis investigates basic properties of the interpoint distances be- tween random vectors drawn from multinomial distribution. We also describe a possible application to testing sparse observations, i.e., a setup with small number of observations and large number of categories, where the classical χ2 -test cannot be recommended. As an alternative, utilizing the multinomial interpoint distances, we will present the test statistic proposed by Biswas and Ghosh (2014). 1
|
|
|
Score tests in contingency tables
Jex, Martin ; Omelka, Marek (advisor) ; Kulich, Michal (referee)
The thesis deals with testing of hypotheses in multinomial distribution. It utilizes two approaches, Pearson's approach known as the of goodness of fit test and the approach stemming from theory of maximum likelihood. The thesis presents derivations of tests based on maximum likelihood. Both approaches are used on the multinomial distribution and for both cases with and without nuisance parameters. The links between both approaches are presented as well. Furthermore both approaches are illustrated on real data to facilitate better understanding of the discussed problems. 1
|
|
|
Confidence Intervals for Quantiles
Horejšová, Markéta ; Kulich, Michal (advisor) ; Hlávka, Zdeněk (referee)
In this thesis, various construction methods for simultaneous confidence intervals for quantiles are explained. Among nonparametric approaches, a special emphasis is dedicated to a recent method based on a multinomial distribution for calculating the overall confidence level of confidence intervals for all quantiles of interest using an efficient recursive algorithm, which is also described. Furthermore, a method based on Kolmogorov-Smirnov statistic or an asymptotic method using empirical distribution function and order statistics for quantile estimate are presented. A special parametric method for several quantiles of a normally distributed population is introduced along with a few of its modifications. Subsequently, a simulation is run to test the real coverage of the described theoretical methods. Powered by TCPDF (www.tcpdf.org)
|
|
|
Confidence intervals for parameters of multinomial distribution
Bárnetová, Kamila ; Anděl, Jiří (advisor) ; Omelka, Marek (referee)
Title: Confidence intervals for parameters of multinomial distribution Author: Kamila Bárnetová Department: Department of Probability and Mathematical Statistics Supervisor: prof. RNDr. Jiří Anděl, DrSc., Department of Probability and Mathematical Statistics Abstract: Confidence intervals for parameters for binomial and multinomial distribution are described in this thesis. These intervals can be used in practice, for exemple- pre-election estimates. The first two chapter are devoted to derivation of these intervals. Simulations and comparison of several selected methods can be found in the last chapter. Based on the simulations, we consider it appropriate, to choose to calculate confidence intervals for parameters of multinomial distribution intervals based on Bonferroniho inequality, or their modifications. These intervals are easy to calculate, while their coverage probability is at least 0.89. Keywords: confidence interval, multinomial distribution, binomial distribution, Bonferroni inequality
|
|
|
Tests in multinomial distribution
Holý, Vladimír ; Anděl, Jiří (advisor) ; Antoch, Jaromír (referee)
In this paper there are at first described classical goodness-of-fit tests - the Pear- son's χ2 test and the log likehood ratio test. The more modern method of testing is the family of statistics based on power divergence which is generalisation of classical statistics. Another type of generalisation is the family of disparity statis- tics which includes beside the family of power divergence also the families BWHD and BWCS. It is demonstrated that all these test statistics have an asymptotic χ2 distribution. In the program R the exact level and exact power can be calculated for individual tests. Hereafter, moments of test statistics can be derived. On the basis of these comparisons there will be shown which test statistics are the most suitable for the goodness-of-fit tests. 1
|
|
|
Retail loan repayment analysis using generalized linear models
Šolc, Michal ; Jarošová, Eva (advisor) ; Forbelská, Marie (referee)
This Diploma thesis concern with generalized linear models and their application in bank practice. Especially to analyze retail loan repayment. First of all we see into theoretical viewpoint of generalized linear models. We shortly try to summarize problems of clasical linear model restrictions and after that we apply to theory on which generalized models are based. We introduce an overview of generalized linear models and after that we concern models, where dependent variable have multinomial and gamma distribution in detail. Main part this thesis is dedicated to data analysis about bank retail loans repayments. In this analysis we use those early mentioned models. We try to create good statistical models on which base the risk ratio of current bank clients could be predicted. The risk ratio is measured by two main indicators, which are: "overdue time" and "overdue amount". For analysis is used statistical software SAS.
|
guest ::
