Analysis interface - Githubissues

Amsterdam-Music-Lab / MUSCLE

An application to easily set up and run online listening experiments for music research.

https://www.amsterdammusiclab.nl/

MIT License

4 stars 1 forks source link

Analysis interface #97

Open BeritJanssen opened 2 years ago

BeritJanssen commented 2 years ago

For preliminary analysis of the result, it may be handy to have an analysis interface which performs some general summary statistics of the data, so not everything has to be downloaded and processed offline at all times.

drikusroor commented 8 months ago

Perhaps we can come up with a list of the 10 most important numbers/data we'd want to see on the results?

BeritJanssen commented 8 months ago

We may split these into two issues:

[ ] as a researcher I want to get summary statistics (mean / variance / standard deviation) about the experiment I have run -> this functionality could be covered with documentation & starter notebooks in R and Python
[ ] as a participant I want to know how well I did, compared to the average performance -> this is already implemented as percentiles in some experiments, we might generalize that.

drikusroor commented 5 months ago

Is this the same as #979 and should we close this issue?

jaburgoyne commented 5 months ago

It isn't really the same, but since that is a big meta-issue, you could wrap this one into it if you think it makes things easier to reason about.

The researcher's view is something we thought about 'outsourcing' to the data science engineers of the Humanities Labs in general: they have good visualisation expertise. On the participant side, it is more MUSCLE specific: another way of describing the issue would be to make it easier to design scoring feedback pages. The trophies in Hooked on Music, for example, work with cutoffs in terms of points or percentages, but it's mostly pretty ad-hoc