Využití časových řad v analýze dat z eye tracking systému

Autor: Bc. Jan ProcházkaVedoucí práce: Doc. Mgr. Jiří Dvorský, Ph.D.


The main aim of the thesis was searching for repeating segments in experimental data which would imply similar ways of solving the problems by different respondents. For this purpose the algorithms working with time series were used, namely the Voting Experts algorithm and the Dynamic Time Warping algorithm. The data was obtained by the eye-tracking method.

One of the minor aims of the thesis was to choose data in a way that would enable further analysis. In the past the Department of Geoinformatics at the Palacký University in Olomouc did some research concerning the differences in the perception between 2D and 3D maps. Some of this data was also used in this thesis.

The program that enabled the implementation of the Voting Experts algorithm as well as the Dynamic Time Warping algorithm was created by Ing. Tomáš Kocyan. Since the program requires the input data to be in 1D format, it was also necessary to create the method that would enable the application of the program over the 2D data obtained from eye-tracking system. For this purpose the scripts were created in the MATLAB, which first convert the data into a suitable form and then make the segmentation of input data. All of such segments were then mutually compared and their similarity henceforth calculated.

© 2013 Jan Procházka, Katedra geoinformatiky, Přírodovědecká fakulta, Univerzita Palackého v Olomouci, honzaproch@centrum.cz