|
|
Automatic sleep scoring using polysomnographic data
Vávrová, Eva ; Potočňák, Tomáš (referee) ; Ronzhina, Marina (advisor)
The thesis is focused on analysis of polysomnographic signals based on extraction of chosen parameters in time, frequency and time-frequency domain. The parameters are acquired from 30 seconds long segments of EEG, EMG and EOG signals recorded during different sleep stages. The parameters used for automatic classification of sleep stages are selected according to statistical analysis. The classification is realized by artificial neural networks, k-NN classifier and linear discriminant analysis. The program with a graphical user interface was created using Matlab.
|
|
|
Building predictive models
ZABLOUDIL, Jakub
This mater thesis is focused on building predictive models. Their fundamental task is to provide an early-warning system, giving information about potential enterprise bankruptcy. The main essence and aim of the thesis is to create multivariate classification models by using discriminant analysis and logistic regression. Emphasis is put on their predictive accuracy, which is assessed for period of three years before bankruptcy declaration. Attempts to optimize classification thresholds in order to increase the initial accuracy are also made. Evaluating classification reliability of several existing models and performing profile analysis assessing predictive ability of univariate ratios were accomplished as well.
|
|
|
Credit Risk Measurement in Manufacturing Industry Companies in the Czech Republic
Karas, Michal ; Dohnal, Mirko (referee) ; Hrvolová, Božena (referee) ; Myšková, Renáta (referee) ; Režňáková, Mária (advisor)
The purpose of this doctoral thesis is to create a new bankruptcy prediction model and also to design how to use this model for the purposes of credit risk measuring. The starting-point of this work is the analysis of traditional bankruptcy models. It was found out that the traditional bankruptcy model are not enough effective in the current economic conditions and it is necessary to create a new ones. Based on the identified deficiencies of the traditional models a set of two new model series was created. The first series of the created models is based on the use of parametric methods, and the second one is based on the use of newer nonparametric approach. Moreover, a set of factors which are able to identify an imminent bankruptcy was analyzed. It was found, that significant signs of imminent bankruptcy can be identified even five years before the bankruptcy occurs. Based on these findings a new model was created. This model incorporates variables of static and even dynamic character for bankruptcy prediction purposes. The overall classification accuracy of this model is 92.27% of correctly classified active companies and 95.65% of correctly classified bankrupt companies.
|
|
|
Statistical Classification Methods
Barvenčík, Oldřich ; Žák, Libor (referee) ; Michálek, Jaroslav (advisor)
The thesis deals with selected classification methods. The thesis describes the basis of cluster analysis, discriminant analysis and theory of classification trees. The usage is demonstrated by classification of simulated data, the calculation is made in the program STATISTICA. In practical part of the thesis there is the comparison of the methods for classification of real data files of various extent. Classification methods are used for solving of the real task – prediction of air pollution based of the weather forecast.
|
|
| |
|
|
Use of value engineering elements for realization of building in the municipality
Chytková, Markéta ; Hanák,, Tomáš (referee) ; Puchýř, Bohumil (advisor)
The aim of my thesis Use of value engineering elements for the realization of buildings in the municipality is to compare the evaluation methods of procurement in municipality and evaluation methods using elements of value engineering. A tender for reconstruction Šubertovo square was realized in the city Dobruška. In the thesis is shown way of evaluation of competing offers of the city Dobruška itself in sublimit simplified procedure and evaluation of competing offers by selecting more evaluation criteria and selecting the best offer with the highest degree of effectiveness.
|
|
|
Processing of image sequences from fundus camera
Klimeš, Filip ; Odstrčilík, Jan (referee) ; Kolář, Radim (advisor)
Cílem mé diplomové práce bylo navrhnout metodu analýzy retinálních sekvencí, která bude hodnotit kvalitu jednotlivých snímků. V teoretické části se také zabývám vlastnostmi retinálních sekvencí a způsobem registrace snímků z fundus kamery. V praktické části je implementována metoda hodnocení kvality snímků, která je otestována na reálných retinálních sekvencích a vyhodnocena její úspěšnost. Práce hodnotí i vliv této metody na registraci retinálních snímků.
|
|
|
Discriminant and cluster analysis as a tool for classification of objects
Rynešová, Pavlína ; Löster, Tomáš (advisor) ; Řezanková, Hana (referee)
Cluster and discriminant analysis belong to basic classification methods. Using cluster analysis can be a disordered group of objects organized into several internally homogeneous classes or clusters. Discriminant analysis creates knowledge based on the jurisdiction of existing classes classification rule, which can be then used for classifying units with an unknown group membership. The aim of this thesis is a comparison of discriminant analysis and different methods of cluster analysis. To reflect the distances between objects within each cluster, squeared Euclidean and Mahalanobis distances are used. In total, there are 28 datasets analyzed in this thesis. In case of leaving correlated variables in the set and applying squared Euclidean distance, Ward´s method classified objects into clusters the most successfully (42,0 %). After changing metrics on the Mahalanobis distance, the most successful method has become the furthest neighbor method (37,5 %). After removing highly correlated variables and applying methods with Euclidean metric, Ward´s method was again the most successful in classification of objects (42,0%). From the result implies that cluster analysis is more precise when excluding correlated variables than when leaving them in a dataset. The average result of discriminant analysis for data with correlated variables and also without correlated variables is 88,7 %.
|
|
|
Comparisons of discriminant analysis and classification trees
Dlabač, Jaroslav ; Vilikus, Ondřej (advisor) ; Stecenková, Marina (referee)
This bachelor thesis compares two methods to discrimination and classification of data in multivariate statistics analysis. While discriminant analysis represents the classical statistical method for discrimination and subsequent classification data method, CART is a new procedure in data-minig, which uses artificial intelligence. The first half of this work is devoted to theoretical description and comparison of these two methods. The second half is the demonstration of both methods on practical example. At the end, the results of both methods are compared and evaluated.
|
|
|
Využití NIR spektroskopie při kontrole falšování kozího mléka
Ptáčková, Eva
The aim of the thesis was to find out whether NIR spectroscopy is a suitable method for adulteration control of cow and goat milk. Also to which concentration level the method is still effective was sought. Theoretical part summarizes information about the requirements for goat milk, its composition and processing possibilities. It also contains information about food adulteration and the relevant legislation. Infrared spectroscopy is theoretically described, particularly NIR spectroscopy. The experimental part describes the procedure of qualitative analysis using NIR spectroscopy, the method of evaluation and its results. Different samples of goat milk adulterated with various proportions of cow milk were compared. Results were evaluated using discriminant analysis. NIR spectroscopy has been demonstrated as a method suitable for detecting goat milk adulteration.
|
guest ::
