We have discussed the foundational measures of information theory, and introduced the use of the JIDT toolkit for estimating information-theoretic measures to analyse relationships in complex systems.
Your task is to create an information-theoretic analysis of a data set of your choosing, selecting appropriate measures to answer one or more questions of interest regarding the dynamics of and relationships in that system. You will deliver an in-class presentation, and write a report, describing your approach, the results, discussing implications for the system under study, and critically evaluating your findings. In detail, you need to:
- (Aim/Materials) Identify:
- and describe a suitable data set which you could apply information theory to;
- a set of questions / hypotheses regarding that data set, which information theory could be
used to address. For example, these could regard variability in the data, relationships between variables, higher order (conditional or multivariate) relationships, and in particular how these measures / relationships vary with respect to time / condition, etc.
- (Method) Determine how you need to pre-process the data, and how to apply your information theoretic measures. For example, which measure will you use to answer your question? Which parts of the data set will be analysed? Which estimator is most appropriate, and how to set the parameters for it here?
- (Results) Apply your analysis, and report the results. Don’t forget that your insights may suggest new analyses that would be interesting to perform, and you shouldn’t be afraid to revisit and adjust your questions and then perform additional analyses.
- (Discussion) Provide a discussion and critical evaluation of your findings. In particular, you should include: How can your results be interpreted, and what insights do they give about the system? Are your results statistically significant? Additional aspects to include as critical analysis will vary based on your experiment, but may include for example: how your
analysis could be improved, how the work could be extended in future, whether the findings may add insights to other systems, etc.
A description of each of the above should be contained in both your presentation and report. You can see that the sections are structured as per a typical scientific lab report.
You must include your code in an appendix of the report. (This will not be included in the word count).
There are many potential sources of data which you could analyse here. These include but are not limited to: open data sets identified during your investigations for assessment 1; sources you know from your background domain; open data sets listed on the discussion board, or found via other searching; or generating your own data from simulations of complex systems (like our Ant foraging example in Module 8). It is perfectly fine to reproduce the results from an existing paper on an open data set, provided the task matches the above specification.
Notes on applicable policies:
This assignment will be submitted via TurnItIn, a text-matching tool used to help detect instances of academic dishonesty. For more detail about TurnItIn and the Academic Honesty Policy, please see the Canvas site.
Please see the Canvas site regarding the late submissions policy. No extensions (outside of the formal special consideration process) will be granted for the Presentation component, although submission by video presentation by the due date may be negotiated if you cannot be present.
