Reference test verse selection

[alexcwatt]

Instead of random sorting in MemversesController#test_next_ref, we should probably sort by last tested.

But in these 20 or more questions, there were about 5 verse references that were questioned more than 3 times each, while it could have drawn on other verses instead.

source.

[Nateowami]

Last tested as in "verse last tested" or "reference last tested"? As far as I can tell the latter is not being tracked, but it could be added. The other thing is that some users could have reviewed every verse quite recently, and if only the day of review were tracked, it could be hard to choose verses for users who only have a few. My thought would be to factor in both last test and interval.

[alexcwatt]

We can have an SQL query get next test minus current interval for
references, which are tracked.
On Aug 25, 2014 8:54 AM, "Nateowami" [email protected] wrote:

Last tested as in "verse last tested" or "reference last tested"? As far
as I can tell the latter is not being tracked, but it could be added. The
other thing is that some users could have reviewed every verse quite
recently, and if only the day of review were tracked, it could be hard to
choose verses for users who only have a few. My thought would be to factor
in both last test and interval.

—
Reply to this email directly or view it on GitHub
#293 (comment).

[Nateowami]

That totally makes sense; I don't know why I didn't think of that.

The accuracy test has the same issue (though not as badly I think), but I doubt that'd be as easy to fix (it's currently completely random).