Dissertation theses
Operations Data Science: Machine learning on massive operational data from systems infrastructures
Supervisor
Level
Topic of dissertation thesis
Topic description
System infrastructures including operating systems, networking, web servers, load balancers, application servers, etc. produce massive amounts of operations data during systems runtime. Operations data science aims to improve various machine learning methods to help understand patterns, correlations or similarities from the past systems’ behavior in order to improve operations practice, for example, systems maintenance, changes in systems design or root cause analysis of incidents. This includes predicting systems behaviors when unplanned or unexpected events occur such as connectivity lost, failures of servers or spikes in systems workloads. This research will improve supervised and unsupervised learning methods tailored for problems from operational data analytics and cover a wide spectrum of underlying machine learning tasks including classification, clustering or regression.
Bachelor theses
Economic and technical analysis of heat pump data
Author
Tereza Ehnová
Year
2023
Type
Bachelor thesis
Supervisor
doc. Ing. Tomáš Vitvar, Ph.D.
Reviewers
Ing. Jaroslav Kuchař, Ph.D.
Department
Summary
This bachelor thesis analyzes an air-to-water heat pump system assisted by a solar panel, focusing on its operation, utilization, and necessary performance metrics. The thesis provides an overview of a specific system, analyzing trends in energy consumption and efficiency and identifying reasons for lower efficiency in the summer season. Additionally, the impact of various factors on the system's efficiency and electricity consumption is assessed, with temperature having the greatest influence on efficiency and compressor speed having the most significant impact on energy consumption. Afterwards, the most influential factors were used for predictions. The investment in a heat pump is advantageous compared to alternative investment options, particularly with the grant obtained, and can be comparable to investing in stocks in some cases.
Analysis of Avalanche Data and Ski Touring Safety
Author
Matyáš Thér
Year
2024
Type
Bachelor thesis
Supervisor
doc. Ing. Tomáš Vitvar, Ph.D.
Reviewers
Ing. Jitka Hrabáková, Ph.D.
Department
Summary
The work focuses on the acquisition of data on historical avalanche events in the Austrian federal state of Tyrol, their preprocessing and supplementation with meteorological data in order to create a dataset. The obtained avalanche incident data are analysed and compared with avalanche bulletins published by LWD Tirol. In order to predict the avalanche danger rating, classification is performed using selected traditional methods (random forest, gradient boosting, SVM and logistic regression). Due to the imbalanced distribution of the classes of the target variable, random and ADASYN oversampling is used to train the models. A value of 0.74 is achieved for the weighted F1 score for the random forest model trained on the training set with random oversampling. The main output of the work is a publicly available repository on GitHub with a preprocessed dataset and analysis procedure in Jupyter Notebook format, which allows replication of the procedure and further analysis of the data obtained.
Master theses
Tool for faceted browsing of studying materials
Author
Vladimír Říha
Year
2012
Type
Master thesis
Supervisor
doc. Ing. Tomáš Vitvar, Ph.D.
Reviewers
Ing. Zdeněk Troníček, Ph.D.
Department
Tool for authoring of HTML5 presentations with capabilities of real-time users` interactions
Author
Petr Mikota
Year
2012
Type
Master thesis
Supervisor
doc. Ing. Tomáš Vitvar, Ph.D.
Reviewers
Ing. Zdeněk Troníček, Ph.D.
Department
Tool for improved presentation of studying materials with HTML5 technology.
Author
Vojtěch Smrček
Year
2012
Type
Master thesis
Supervisor
doc. Ing. Tomáš Vitvar, Ph.D.
Reviewers
Ing. Zdeněk Troníček, Ph.D.
Department
Application for editing and review of written exams
Author
Adam Havel
Year
2014
Type
Master thesis
Supervisor
doc. Ing. Tomáš Vitvar, Ph.D.
Reviewers
Ing. Milan Dojčinovski, Ph.D.
Department
Summary
The goal of this thesis is to build a web application that would represent a complete solution to the process of creating, taking and evaluating exams. Apart from providing an overview of the implementation, the text tries to describe in detail the decisions that were made while developing the application and to introduce to the reader the many ideas these decisions were based on.
System for central management and integration of sports reservation systems
Author
Jan Fránek
Year
2016
Type
Master thesis
Supervisor
doc. Ing. Tomáš Vitvar, Ph.D.
Reviewers
Ing. Zdeněk Rybola, Ph.D.
Department
Summary
This thesis is focused on the integration of reservation systems for sport facilities. The work starts with a survey and analysis of the existing reservation systems for sport facilities. Next there is an analysis of integration of reservation systems for hotels. Afterthat I proposed functional requirements, data model, system architecture and interfaces for integration. The next sections describe the system implementation and system testing.
Log Server Analytics
Author
Daniil Fedotov
Year
2022
Type
Master thesis
Supervisor
doc. Ing. Tomáš Vitvar, Ph.D.
Reviewers
Ing. Jaroslav Kuchař, Ph.D.
Department
Summary
Currently, information systems from the infrastructural point of view containing operating systems, web servers, application servers, load balancers, network communication elements produce a huge amount of operations data and write it to logs. Analysis of such logs can help to understand the current state of the system -- for instance, to determine the places, when errors occur. Such a knowledge helps to improve the performance of the system, make it more stable and accessible. Human is not able to read, analyze and process such a number of messages in a reasonable time, so it makes sense to perform a cluster analysis on server log files. Several groups (clusters) of similar log messages will be generated after clustering. This approach allows to significantly reduce the dimension of original log data, and allows to analyze not individual log entries, but groups, which simplifies searching and elimination of problems that have arisen during the server runtime.
This thesis proposes software for server log clustering based on natural language processing and machine-generated text processing techniques and machine learning algorithms, followed by the analysis of clustered log data.
Detection of Dark Patterns on Czech Webshops
Author
Petr Hanzl
Year
2022
Type
Master thesis
Supervisor
doc. Ing. Tomáš Vitvar, Ph.D.
Reviewers
Ing. Karel Klouda, Ph.D.
Department
Summary
This thesis investigates patterns in user interfaces, also known as dark patterns, that force users to do things or make decisions differently than they originally intended. This thesis focuses on the detection of dark patterns used by webshops on the Czech Internet and the detection is done on a large scale.
This thesis builds on research already conducted at Princeton University that investigated dark patterns on English webshops.
Several tools were created to retrieve a significant number of webshops. Also the tools from the conducted research were modified to be applied to the Czech language.
These tools were used to obtain multiple datasets mapping webshops on the Czech Internet and the dark patterns used on them.
It was found that dark patterns are widely used on Czech webshops.
Video lectures indexing service
Author
Jiří Zdvomka
Year
2022
Type
Master thesis
Supervisor
doc. Ing. Tomáš Vitvar, Ph.D.
Reviewers
Ing. Milan Dojčinovski, Ph.D.
Department
Summary
Video learning material indexes are rare and could lead to a better studying experience for students. This thesis aims to implement a web service for video lecture indexing with the help of supplementary materials as user input with focus on FIT CTU courses. The result is a deployed web service capable of video lecture indexing with satisfactory accuracy on the testing dataset, providing API, UI and core module with algorithm as a Python package.
Analysis and Comparison of Application Architecture: Monolith, Microservices and Modular Approach
Author
Martin Skalický
Year
2024
Type
Master thesis
Supervisor
doc. Ing. Tomáš Vitvar, Ph.D.
Reviewers
Ing. Jiří Mlejnek
Department
Summary
Microservices architecture have become a ubiquitous standard in the software industry, where they are often used as a one-size-fits-all solution, disregarding their drawbacks. This thesis aims to analyse the design approaches of monolithic, microservices and modern modular architecture and compare them, while emphasising the frequently overlooked negative consequences of Microservices in favour of Modular Monolith architecture. The author discusses his experience working on microservices projects and the challenges they faced with the architecture in practice. A proof of concept application is developed for each type of architecture mentioned and thoroughly analysed for performance and latency. Finally, the thesis concludes by presenting a methodology for employing the Modular Monolith architecture approach in new projects and how the said architecture can evolve throughout the application lifecycle.