To reverse a quote from Donald Knuth: "Beware of bugs in the above code;
I have only tried it, not proved it correct."

The current formula produces results that behave the way we were wanting
them to behave: the more ratings there are, the closer they get to their
raw mean as opposed to 3.0. <https://wiki.ubuntu.com/SoftwareCenter#top-
rated> It's quite possible that it is not exactly correct, and if making
it exactly correct does not involve massive computation, certainly we
should fix it. So, thanks for posting that question.

I disagree, though, that "ratings are fundamentally ordinal data"
necessarily leads to "a mean doesn't make much sense". For example, shoe
sizes and off-the-rack dress sizes are ordinal, but it still makes sense
to talk about whether the average for either of those is increasing over
time. One "Excellent" rating plus one "Terrible" rating does not lead to
"Mediocre", but rather "We can't say with any confidence that this is
better or worse than mediocre".

-- 
You received this bug notification because you are a member of Ubuntu
Bugs, which is subscribed to Ubuntu.
https://bugs.launchpad.net/bugs/894468

Title:
  Statistics algorithm for sorting ratings looks fishy

To manage notifications about this bug go to:
https://bugs.launchpad.net/ubuntu/+source/software-center/+bug/894468/+subscriptions

-- 
ubuntu-bugs mailing list
[email protected]
https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs

Reply via email to