|
|
Use of Physiological Data to Analyze and Improve the User Experience
Štefeková, Nina ; Beran, Vítězslav (referee) ; Materna, Zdeněk (advisor)
This thesis is concerned with the research of the physiological symptoms of stress, the design and execution of an experiment that can induce short-term stress. Subsequently, the goal was to create a dataset from the experimentally obtained data and use it to train a machine learning model. Stress detection by such a model can be used to analyze the user experience. The proposed experiment uses a combination of stimuli, consisting of a relaxation part, a Stroop test and a web game that deliberately uses an unfriendly user interface. During it, data was recorded using an Empatica E4 device and then processed by an implemented application into the final dataset. A machine learning model that can detect short-term stress was then trained and evaluated. The KNN algorithm model evaluated by cross-validation achieves an accuracy of 84\% when the subject of the is known prior to the model. For an unknown subject, it is on average 78\%. The thesis provides this model and the resulting dataset, for further use. These results show that short-term stress detection is more challenging without prior knowledge of the subject.
|
|
|
Vliv chovatelských zásahů na produkční parametry chovu keříčkovce červenolemého (Clarias gariepinus).
Doule, Jan
The aim of the work was to evaluate, based on a feeding test, the influence of commonly performed husbandry interventions on the production results of the breeding of the African catfish (Clarias gariepinus, Burchell 1822). The longterm feeding test was performed on the RAS recirculation system and the fish were reared in aquariums (180 l). The test itself lasted 56 days and was preceded by an 18-day adaptation of the fish to new conditions. Husbandry interventions (control weighing, refishing, mixing and sorting of fish) were applied to the fish at 14-day intervals, and each variant of the husbandry intervention was tested in two repetitions. The results were statistically evaluated by the Kruskal-Wallis test. The Control, Overfishing, Mixing variant achieved the same production results (FCR/SGR 0.25). The Sorting variant showed better production results (FCR/SGR 0.23). In this test, it was proven that the selected breeding interventions do not negatively affect the production results of the breeding of the African catfish. The partial goal of the work was to evaluate how acute stress affects the quality of the internal environment of this species. A 2-day experiment was carried out. In the experiment, the fish were divided into 2 group variants Control and Stress. For the Stress variant, the husbandry intervention of overfishing was applied for 10 minutes (induction of acute stress). Subsequently, blood was collected from 10 fish from both groups in the time interval of 0, 1, 3, 6, 12 and 24 hours of exposure to the stressor. The results of the hematological and biochemical parameters of the stressed fish were compared with the control group. As a result of the stress reaction in the Stress variant, there was first an increase in the values of the Hk, Hb and Ery parameters, then a decrease in them and, after the stressor had subsided, a return to the original range before the stressor (after 24 hours). In the Stress variant fish, there was an increase in glucose content values from 6 to 24 hours compared to the Control variant, which was apparently caused by glycogenolysis in the muscles and liver.
|
guest ::
