openworm
Create rank sort summary
The goal is to summarize how a worm's features sort relative to other worms. As a simple example, one of the outputs from this "function" might be that the worm's width was in the 50th percentile. Alternatively, one might find that a simulated worm is in the 0th percentile for width. This would suggest whatever is controlling width in the simulation is not appropriate.
For now the steps to compute this would be:
- Get histograms for "control" worms
- Get a histogram for the example worm
- Compute basic statistics - see Issue #76 , although we don't really need anything fancy, just a mean or median would be fine
- Sort these statistics for all control worms, and find where the example worm fits
I'm not exactly sure of the output, perhaps an iterable of tuples. It would probably be best to encapsulate in an object for display and for summarizing. An example summary graph might be the distribution (histogram) of rank orders. One could also return the values sorted (via method or property/method) from best (closest to 50%) to worse (0 or 100%).
For example: width is at 50% length is at 80% range is at 80% Then just build a histogram of percentiles
** The goal ** is to provide an easy metric for how the example compares to the group. This could be run on the "control" group as well, since not every control example can be in the 50th percentile (someone has to be the largest or smallest).
A related task: #109
