search

pmelchior / des-exp-checker

Crowdsourced image quality control for the Dark Energy Survey
MIT License
9 stars 2 forks source link

Statistics and "false" flag? #57

Closed kadrlica closed 10 years ago

kadrlica commented 10 years ago

It looks like the statistics are not picking up issues tagged with "False". Is this the intent?

I'm also not sure how to access the false instances through the API. By default, shouldn't all (both false and positive) instance be returned? I've tried: api.php?release=Y1A1&problem=Satellite api.php?release=Y1A1&problem=Satellite&false api.php?release=Y1A1&problem=Satellite&false=true but all of these return a blank list (I know I've flagged at ~5 false satellite tracks).

kadrlica commented 10 years ago

The fix allows the user to access the "false" instances with: api.php?release=Y1A1&problem=Satellite&false

However, the false instances are still not displayed on the statistics page and the generic query returns only non-false instances. Personally, I think it would make the most sense for the statistics page to display both false and non-false instances and for there to be some mechanism in the api to create a file consisting of both.

pmelchior commented 10 years ago

For the statistics page, I think it is better to only show that actual errors, but the request for the API can be easily accommodated.

kadrlica commented 10 years ago

I think that the statistics page should track both positive and negative instances of problems at the same time. Right now I think false satellite trails are one of the biggest issues, but that is not reflected by the statistics page. Are images with "false" problems in them making it in the the "Ok" slice of the pie wedge?

pmelchior commented 10 years ago

No, OK is only if there is no problem.

One could add the statistics for false positives to each of the donuts. Would not be too hard. Probably doesn't need to be available today...

kadrlica commented 10 years ago

Yeah, that sounds ideal.