|
Nonperforming Loans in China: Do We (Still) Need to Worry?
Sočuvková, Alžbeta ; Geršl, Adam (advisor) ; Lopušník, Ondřej (referee)
This thesis overviews and discusses the issue of nonperforming loans in general and in China, and financial sector development from the beginning of Deng Xiaoping's reform era in 1978. The focus is put on the underlying institutional framework as it is found to be one of key contributing factors of NPLs in China. Utilizing a sample of 23 countries over the period 2000-2009, we estimate the impact of macroeconomic determinants on nonperforming loans. One of the regressors, output gap, is estimated via Kalman filtering technique based on the multivariate unobserved component (MUC) method. Consequently, out-of-sample method forecasts for five subgroups of countries, thus providing different benchmarks for China, are compared with official data. The conclusion argues the NPLs ratio to be understated in official statistics, especially towards the end of the series, which might be a sign of increasing macroeconomic instability in China.
|
|
The optimal route search under several criteria
Vodička, Jan ; Fiala, Jiří (advisor) ; Kolman, Petr (referee)
Common methods of optimal route search are described in the thesis. Also, existing procedures used in problematics of transportation networks are mentioned. A nonlinear price function is proposed, which enables the user to specify their own preferences of any criteria (length, time, cost, etc.). Moreover, necessary modifications of common optimal route searching algorithms are presented. A method for propagation of maneouvres of arbitrary length into the transportation network is shown. This method enlarges nodes and vertices sets by amount linearly proportional to number of nodes involved in maneouvre set. Also, a general method diminishing the number of edges, which have to be encountered during optimal raute search process, is proposed. Precalculated edges sets are used to gain this goal. Proposed thesis contains three methods solving specific aspects of optimal route search in transportation networks. While the first method, when applied in practice, can bring limitations of processable data size, the other two procedures forrn the basis of a navigation system. Powered by TCPDF (www.tcpdf.org)
|
|
Diagnostics for Robust Regression: Linear Versus Nonlinear Model
Kalina, Jan
Robust statistical methods represent important tools for estimating parameters in linear as well as nonlinear econometric models. In contrary to the least squares, they do not suffer from vulnerability to the presence of outlying measurements in the data. Nevertheless, they need to be accompanied by diagnostic tools for verifying their assumptions. In this paper, we propose the asymptotic Goldfeld-Quandt test for the regression median. It allows to formulate a natural procedure for models with heteroscedastic disturbances, which is again based on the regression median. Further, we pay attention to nonlinear regression model. We focus on the nonlinear least weighted squares estimator, which is one of recently proposed robust estimators of parameters in a nonlinear regression. We study residuals of the estimator and use a numerical simulation to reveal that they can be severely heteroscedastic also for data generated from a model with homoscedastic disturbances. Thus, we give a warning that standard residuals of the robust nonlinear estimator may produce misleading results if used for the standard diagnostic tools
|
|
On Exact Heteroscedasticity Testing for Robust Regression
Kalina, Jan ; Peštová, Barbora
The paper is devoted to the least weighted squares estimator, which is one of highly robust estimators for the linear regression model. Novel permutation tests of heteroscedasticity are proposed. Also the asymptotic behavior of the permutation test statistics of the Goldfeld-Quandt and Breusch-Pagan tests is investigated. A numerical experiment on real economic data is presented, which also shows how to perform a robust prediction model under heteroscedasticity.
|
|
Regression methods for statistical analysis of spatial data
Klimprová, Lucie ; Šerák, Petr (referee) ; Michálek, Jaroslav (advisor)
Kriging techniques are regression methods used for evaluation of continuous spatial processes. If the covariance structure of process is unknown, then it's necessary to estimate it from the data. The first part of this Master's thesis is devoted to description the kriging method and to estimate of a variogram fuction, which describes the covariance structure of considered process. The second part includes the implementation of kriging method in MATLAB for simulated and real data.
|
|
Modelling, parameter estimation, optimisation and control of transport and reaction processes in bioreactors.
ŠTUMBAUER, Václav
With the significant potential of microalgae as a major biofuel source of the future, a considerable scientific attention is attracted towards the field of biotechnology and bioprocess engineering. Nevertheless the current photobioreactor (PBR) design methods are still too empirical. With this work I would like to promote the idea of designing a production system, such as a PBR, completely \emph{in silico}, thus allowing for the in silico optimization and optimal control determination. The thesis deals with the PBR modeling and simulation. It addresses two crucial issues in the current state-of-the-art PBR modeling. The first issue relevant to the deficiency of the currently available models - the incorrect or insufficient treatment of either the transport process modeling, the reaction modeling or the coupling between these two models. A correct treatment of both the transport and the reaction phenomena is proposed in the thesis - in the form of a unified modeling framework consisting of three interconnected parts - (i) the state system, (ii) the fluid-dynamic model and (iii) optimal control determination. The proposed model structure allows prediction of the PBR performance with respect to the modelled PBR size, geometry, operating conditions or a particular microalgae strain. The proposed unified modeling approach is applied to the case of the Couette-Taylor photobioreactor (CTBR) where it is used for the optimal control solution. The PBR represents a complex multiscale problem and especially in the case of the production scale systems, the associated computational costs are paramount. This is the second crucial issue addressed in the thesis. With respect to the computational complexity, the fluid dynamics simulation is the most costly part of the PBR simulation. To model the fluid flow with the classical CFD (Computational Fluid Dynamics) methods inside a production scale PBR leads to an enormous grid size. This usually requires a parallel implementation of the solver but in the parallelization of the classical methods lies another relevant issue - that of the amount of data the individual nodes must interchange with each other. The thesis addresses the performance relevant issues by proposing and evaluation alternative approaches to the fluid flow simulation. These approaches are more suitable to the parallel implementation than the classical methods because of their rather local character in comparison to the classical methods - namely the Lattice Boltzmann Method (LBM) for fluid flow, which is the primary focus of the thesis in this regard and alternatively also the discrete random walk based method (DRW). As the outcome of the thesis I have developed and validated a new Lagrangian general modeling approach to the transport and reaction processes in PBR - a framework based on the Lattice Boltzmann method (LBM) and the model of the Photosynthetic Factory (PSF) that models correctly the transport and reaction processes and their coupling. Further I have implemented a software prototype based on the proposed modeling approach and validated this prototype on the case of the Coutte-Taylor PBR. I have also demonstrated that the modeling approach has a significant potential from the computational costs point of view by implementing and validating the software prototype on the parallel architecture of CUDA (Compute Unified Device Architecture). The current parallel implementation is approximately 20 times faster than the unparallized one and decreases thus significantly the iteration cycle of the PBR design process.
|
|
Requirements life cycle for software development in the bank
Bambas, Pavel ; Pour, Jan (advisor) ; Koten, Jiří (referee)
This thesis deals with the introductory part of the requirements life cycle for software development in the bank, paying particular attention to the following: data collection methods; size estimation; prioritization; and scheduling. This work will compare different approaches appropriate for rigorous and agile methodologies. Analysis of the findings from practice in a small bank should help college graduates who find initial employment in the position of Business Analyst / IT Coordinator with only a basic orientation in this topic.
|
|
Quantitative structure-antimycobacterial acitivity relationships in the group of potential antituberculotics
Doležal, Rafael ; Waisser, Karel (advisor) ; Lehotay, Jozef (referee) ; Ponec, Robert (referee)
DOLEŽAL Rafael: Quantitative relationships between the structure and antimycobacterial activity of selected potential antituberculotics. Hradec Králové: Faculty of Pharmacy in Hradec Králové, Charles University in Prague, 2008, 274 pp. Ph.D. Thesis. The Ph.D. thesis deals with analysis of quantitative relationships between the structure and antimycobacterial activity (QSAR) of four structural classes of potential antituberculotics - derivatives of N-phenylcarbamic acid esters, derivatives of 1-phenyl-1H-tetrazole-5-thiol, derivatives of salicylamides and derivatives of salicylthioamides. In total, 600 computer models of different compounds were created. 204 of them, after conformational analysis in HyperChem 7.52 and geometric optimization with the B3LYP/6-31G* DFT method in Gaussian 03, provided data for calculation of several types of descriptors. Attention was turned especially to local, quantum-chemical descriptors of electronic properties, such as superdelocalizability, index of frontier electron density and Mulliken electric charge. The QSAR analyses are based on searching for statistically significant correlations between the antimycobacterial activities in vitro and a matrix of various descriptors by means of multiple linear regression (MLR) with the use of a selection algorithm, by the...
|
| |
|
What is the price elasticity of demand for owner-occupied housing for region of Pardubice?
Skalická, Eliška ; Rod, Aleš (advisor) ; Matějka, Marek (referee)
This bachelor thesis examines an assumption that a demand for owner-occupied housing is price-inelastic in the district of Pardubice. The data for the period from 2008 to 2012 are used in the thesis. The test has proved the demand for housing is truly price-inelastic; it means that 1% increase in housing prices causes a decrease by 0,85 % of demands. Some control variables are also added into the model. This expanded model estimates coefficients in range from 0,72 to 0,96 in absolute values.
|
guest ::
