Page MenuHomePhabricator

Summary statistics for /inspect
Closed, ResolvedPublic

Description

Set up /inspect to just be summary statistics so you keep double-blind review, and use file manipulation commands on the shell command line if you need to.

Event Timeline

prnk28 created this task.Jun 30 2016, 4:58 PM
Restricted Application added subscribers: Zppix, Aklapper. · View Herald TranscriptJun 30 2016, 4:58 PM

@prnk28 also please consider printing a table of dates for {all, q, a, e,
o, t, d} X {oldest modification date, newest date, mean date, median date,
standard deviation days, how many have different creation and modification
dates}

@prnk28 I am also thinking we should have coverage statistics here: once you make one of those 7x6 tables (from my comment T139079#2418938 above) you can print one for each of the WP:BACKLOG categories.

Jsalsman triaged this task as High priority.Jul 2 2016, 11:01 PM
Jsalsman updated the task description. (Show Details)

@prnk28 HAVE YOU NOTICED HOW FAR AHEAD OF SCHEDULE THE REFACTORING GOT US? :D

The only thing we need to pick back up off of the redesign floor is tokens for reviewers so you can run reputation scores on them. Maybe if GET /inspect is given an optional argument like /inspect[?has=string] then it can search for files which contain those strings to make a 7x6 statistics table? Then you can just use the category names for strings and not have to list them all in your code!

@prnk28 ...the point being, that the ratio of reviewer identification token strings occurring in file sets with -e instead of -o (and something mumble mumble about -t we will figure out) is the same as the reviewer's agreement score, and, well, you probably need more statistics to figure out the reputation score as we originally designed it from that, but I'm pretty sure that can be part of GET /inspect?has=RevewByWikiGnome543 somehow.

Jsalsman claimed this task.Jul 3 2016, 6:01 AM

I am working on this one....

Jsalsman updated the task description. (Show Details)Jul 3 2016, 6:04 AM

@prnk28 also please consider printing a table of dates for {all, q, a, e,
o, t, d} X {oldest modification date, newest date, mean date, median date,
standard deviation days, how many have different creation and modification
dates}

What does the last column mean? We aren't modifying files once they've been created right?