|
|
Generating Structured Test Data
Olšák, Ondřej ; Holík, Lukáš (referee) ; Smrčka, Aleš (advisor)
The goal of the bachelor's thesis is to create a tool for generating files with structure data content. The purpose of these files is to be used as test data conforming to testing of program input space. This thesis focuses on tree-structured data. The tool integrates tools implemented previously within Testos framework for generating test data in order to satisfy user-defined coverage criterion. The tool is able to generate a set of files in JSON or XML format containing test data satisfying ECC, BCC, or PWC coverage criterion.
|
|
|
Model of geometry of micro-lattice structures for finite element method
Javorský, Dominik ; Maňas, Pavel (referee) ; Červinek, Ondřej (advisor)
The growing availability of additive manufacturing technology, such as SLM, enables the creation of innovative structural designs. An example of these are complex lattice structures which are used because of their excellent mechanical properties relative to weight. One significant challenge lies in production tolerances. These are not crucial for the majority of components. However, when dealing with lattice structures and their thin-walled geometry, manufacturing tolerances lead to substantial deviations in mechanical properties. The process of designing such structures and achieving results comparable to the experiment requires the use of non-standard methods. These methods are time-consuming and costly for obtaining real geometry. The real geometry is then used for numerical simulations based on the FEM principle. This thesis focuses on the impact of geometric imperfections occurring in BCC-type structures made of stainless steel 316L. During the solution, the real geometry was obtained through optical digitization using the ATOS Triple Scan scanner. Dynamic drop-weight tests were also conducted, and the obtained results were used to modify the geometry model in combination with the acquired real geometry. The aim was to minimize the deviation between experimental and numerical simulation results below 5%. The knowledge gained from this process was then applied in simulations investigating the impact of geometric imperfections. Deviations up to 30% were observed in simulations investigating the impact of geometric imperfections. These deviations can be minimized by incorporating the knowledge of real geometry into the design. The results also help determine the diameter range within which including geometric imperfections in the design is irrelevant. Furthermore, a significant impact of the node geometry on the results of numerical simulations was observed. This knowledge brought the values closer to the experimental data. Another important contribution of this work is the simplified geometry model. This model will enable the study of the impact of additional imperfections in a range that was previously unattainable.
|
|
|
Attention Based High Resolution Image Classification
HEINDL, Dominik
Modern digital images, especially in the field of medicine, have extremely high resolutions. Current state-of-the-art image recognition techniques, like Convolutional Neural Networks, cannot handle such high dimensional inputs. In this thesis I compared the standard approach ofclassifying images by downscaling them with an attention-based Multiple Instance Learning approach where the original image is split up into several smaller patches and low dimensional embeddings are calculated for each patch by a Convolutional Neural Network. All low dimensional embeddings are then again processed in a MIL fashion, where attention-pooling is used to determine the importance of each patch. The data set for this thesis consisted of ultra high resolution histological slides of human skin which were classified to contain Basal Cell Carcinoma or not.
|
|
|
Empirical analysis of the effectiveness of English Premier League football clubs in the 2019/20 season
Smola, Vojtěch ; Gregor, Martin (advisor) ; Vávra, Jan (referee)
This paper focuses on an empirical analysis of the efficiency of English Premier League football clubs. Scholarly papers, that have focused on this topic previously, have reached different conclusions. The aim of this thesis is to show whether clubs manage to spend their finances efficiently or not, and furthermore to identify the most efficient teams. For this purpose, we have used data from 2019/20 season. We focus on club expenditure (players' and other staff salaries, advertising, new player purchases) and income (tickets, broadcasting rights, sponsors and how many points they earned in this season. We will also take note of the capacity of the stadium. The method used was data envelopment analysis with the CCR model, which is input-oriented, and the BCC model, which uses variable returns to scale, as they are the most appropriate for this purpose. The results show that big clubs or teams successful internationally are not more efficient than other league teams. The inefficiency is primarily due to poor use of resources, but also by the global pandemic of Covid-19.
|
|
|
Generating Structured Test Data
Olšák, Ondřej ; Holík, Lukáš (referee) ; Smrčka, Aleš (advisor)
The goal of the bachelor's thesis is to create a tool for generating files with structure data content. The purpose of these files is to be used as test data conforming to testing of program input space. This thesis focuses on tree-structured data. The tool integrates tools implemented previously within Testos framework for generating test data in order to satisfy user-defined coverage criterion. The tool is able to generate a set of files in JSON or XML format containing test data satisfying ECC, BCC, or PWC coverage criterion.
|
|
|
Evaluating of Efficiency of Football Clubs in Premier League by Data Envelopment Analysis
Konečný, David ; Šíma, Jan (advisor) ; Voráček, Josef (referee)
Title: Evaluating of efficiency of Football Clubs in Premier League by Data Envelopment Analysis. Goals: The aim of the thesis is to identify the effectiveness of football clubs in the Premier League in the season 2016/2017. In the post optimization analysis evaluate, which observed clubs have been effective in transforming inputs into outputs, and which clubs have some deficiencies in this transformation. Methods: In the thesis for efficiency research, the Data Envelopment Analysis (DEA) data analysis method is used to evaluate the effectiveness of individual clubs in the Premier League. DEA determines which units are effective and what are the deviations from the effective frontier for the units that are inefficient. The measurements are made by an input- oriented CCR model and a BCC model. The CCR model assumes constant returns to scale and BCC considers variable returns to scale. Results: The result section identifies the productive efficiency of individual football clubs in the Premier League in the season 2016/2017. The effective frontier reached a total of 7 clubs in both CCR and BCC models. The average efficiency in the CCR model is 87 %. In the BCC model, the average efficiency is 91 %. As a result, the Premier League as a competition is highly efficient. Key words: data envelopment...
|
|
|
Efficiency of football clubs in Premier League
Topinka, Jiří ; Šíma, Jan (advisor) ; Voráček, Josef (referee)
Title: Efficiency of football clubs in Premier League Objectives: The aim of this diploma thesis is to investigate the technical efficiency of Premier League clubs in the 2015/2016 season using a Data Envelopment Analysis approach. Determine which clubs worked efficiently and identify weaknesses of individual teams. Methods: The efficiency of each Premier League club is analysed by Data envelopment analysis (DEA). The efficiency is computed for input oriented CCR model (constant returns to scale) and BCC model (variable returns to scale). Results: The technical efficiency of each club in the 2015/2016 Premier League season is evaluated in the practical part of this thesis, with eight clubs achieving maximum efficiency for both CCR and BCC models. That indicates that the competition as a whole is highly effective. Keywords: Premier League, technical efficiency data envelopment analysis, CCR, BCC
|
|
|
DEA models and their application in the measurement of the effectiveness of emergency medical services in different regions of the Czech Republic
Bártíková, Ivana ; Borovička, Adam (advisor) ; Kobzareva, Maria (referee)
Data envelopment analysis (DEA) is a new method of multi-criteria decisions and is used to measurement effectiveness of various processes. This method will be also main subject of this thesis. There will be introduced functional principle of this method including formulations of all mathematical models in first part of thesis. Main differences between measurement of effectiveness in public and private sector will be also described. Analysis of the effectiveness of emergency medical services in different regions of the Czech Republic will be done in second part of this thesis. The data will be applied to both classic and modified models. At the end of the thesis the final results will be compared with indicators of living standard in Czech Republic and optimal solution in different regions will be proposed.
|
|
|
Multicriteria analysis of the Human Development Index
Janů, Jakub ; Střížová, Vlasta (advisor) ; Jablonský, Josef (referee)
Diploma thesis describes the problem of quantitative expression of the quality of life. For this purpose is used one of the most widespread indicators of quality of life - Human Development Index. This paper describes its basic characteristics, method of calculation, its advantages and drawbacks. One of the objectives of this paper is the elimination of these shortcomings by alternative method of calculation. The methods of multi-criteria evaluation of alternatives and data envelopment analysis models are applied for these purposes. One part of this paper is a theoretical description of the tasks of multi-criteria decision making and their classification into the methods for multi-criteria evaluation of alternatives and into the methods of multi-criteria linear programming. In this paper are used methods WSA, TOPSIS and PROMETHEE, based on the definition of the theoretical characteristics of the methods of multi-criteria evaluation of alternatives. These methods are applied on the source data, obtained from a model of the human development index. After calculation, the results obtained are analyzed and written into the conclusion. Another alternative approach to obtain the level of human development is to evaluate the degree of efficiency of countries using data envelopment analysis models in comparison with the human development index. For these purposes is written the theoretical background, which defines the basic types of data envelopment analysis models. These theoretical findings serve as the basis for the practical analysis by the BCC output oriented model and by the corresponding model of super efficiency. The results of these calculations are analyzed and subsequently written into the final conclusion.
|
|
|
Methods for ranking efficient units in data envelopment analysis
Gőtz, Ondřej ; Jablonský, Josef (advisor) ; Chrobok, Viktor (referee)
This thesis focuses on the theory of models, data envelopment analysis and subsequent methods in ranking of obtained efficient units. The thesis is divided into three main parts. The first part describes the basic theoretical background of data envelopment analysis for different situations of inputs and outputs. It also contains basic information on specific models of CCR, BCC and SBM, including illustrative examples of their function. In the second part these models are adapted for use in the overall layout of production units which are the models of super efficiency and cross efficiency. Both methods include the description of the basic principles of their functioning. The aim is to compare the results of the various types of ranking using a practical example. This is included in the last part of the thesis. This chapter compares the results obtained using the models mentioned above. Data of the S & L branches were used as the input values for this work.
|
guest ::
