Computational superstition

From Master Projects
Jump to: navigation, search


About Computational superstition

  • This project has not yet been fulfilled.
  • This project fits in the following Bachelor programs: {{#arraymap:|, |xXx|bachelorproject within::xXx|,}}
  • This project fits in the following masterareas: {{#arraymap:Computational Intelligence and Selforganisation, Technical Artificial Intelligence|, |xXx|project within::xXx|,}}


Description

The project involves two main tracks: Firstly, various datasets are to be downloaded from public sources, and various superstitions - using the relevant data set - are to be analysed using the data. For example, are there significantly more accidents on a Friday that happens to be the 13th day of a month, etc.? Can possible correlations be explained in some rational manner (literature will be helpful in these discussions)? Secondly, can we model how superstitious behaviour is born? For example, if we build a model with only several data instances at hand, we will certainly overfit the data and might build rules that do not generalise well ("When I arrived at the Netherlands it was raining, and now I expect rain every single morning"). More data usually eliminates wrong "branches" of models, but humans often keep holding superstitious thoughts. Without going very deep into the social aspect, can we somehow model/simulate an effect where wrong branches of a model remain even with large numbers of data instances? What is the dynamics of such a process? The project is focused on the computer science/modelling aspect of superstition, but it is recommended that it is supported by findings from social sciences/psychology.