Time-series mining in a psychological domain
Vladimir Kurbalija, Charlotte von Bernstorff, Hans-Dieter Burkhard, Jens Nachtwei, Mirjana Drenovak Ivanović, Lidija Fodor · 2012
Analysis of time-series became an inevitable tool in applied settings, such as stock market analysis, process and quality control, observation of natural phenomena, medical treatments, and in the behavioral science, such as psychological research. In this paper, we utilize a new kind of a tool set for time-series analysis (FAP, developed at Department of Mathematics and Informatics, University of Novi Sad) on behavioral data gained from a specific experimental lab system, a so called Socially Augmented Microworld with three human participants (developed by informatics and psychologists for Human Factors Research at Humboldt University Berlin). On the basis of these data (logfiles) we extracted three types of time-series and generated distance matrices using three kinds of time-series similarity measures. Finally, the clustering of generated distance matrices produced dendrograms which serve as the basis for a deeper analysis of human behavior. The outcome of this analysis is two-fold: (a) it allows to select the most suitable similarity measure for this domain of experimental research and (b) these results can serve as a basis for the development of artificial agents, which may in turn replace the human participants in the experiment.