-
Notifications
You must be signed in to change notification settings - Fork 26
Input from PyAstro 2016
Daniela Huppenkothen edited this page Apr 11, 2016
·
1 revision
The organizers of Python in Astronomy 2016 used Entrofy to do their own participant selection. Here's a rough summary of their process and their suggestions for improvement, as transcribed by Daniela.
- used ipython notebook instead of the GUI
- might want to have an ipython tutorial on how to do the participant selection in a more programmatic way, rather than from the GUI
- in notebook, could use notebook sliders and bokeh?
- in web interface: only the selection is highlighted, if there are few participants selected out of a large pool, this will not be obvious
- need a set of guidelines on how to format the google form so that it gives the information required: e.g. minimize free-form text, make drop-down lists or item sets, form needs to determine set
- want to be able to assign numbers to categories instead of strings
- want to have an easy way of taking floats and making them into ranges (want the user to decide on ranges based on the float value)
- more control over continuous ranges
- in general, make the interface easier to use
- make it easy to adjust weights and categories on the fly
- would like to compare distributions from different outputs
- would want figures also in python layer, not just JavaScript interface
- maybe an object that defines the input of the entrofy procedure: e.g. change weights as attributes, re-run easily via something like
entrofy.run
- output as a table or something similar, such that inputs can be easily copied, small changes made and then run on the new copy
- result could be object that yields the score, the set of selected participants, etc.
- it should be easier to do the second layer after participants have been accepted, but others have declined
- awkward to pass in the pre-selected result
- would be useful to have an easy way to be able to remove someone from the table and then select more, for example via a column "skip