[LON-CAPA-users] Running problem statistics
Guy Albertelli II
Mon, 16 Apr 2007 15:21:30 -0400 (EDT)
> It has been suggested that I run complete statistics on a course at the end
> in order to generate usage data for the problems in the course. I have a
> couple of questions about the process:
> 1. When you start the process, is it necessary to leave that window open
> while the statistics are being generated (apparently about a 30 minute
> 2. Can you go to a role in another course to run those statistics while the
> first set is being run?
> 3. Do you need to stay logged in?
> 4. Suppose you run statistics in the middle of a course, then
> again at the end. Lets say on a particular problem only half the
> students had worked it when you ran the first time, and all students
> had worked it when you ran the second time. Does the second
> statistics run only cover those student tries made since the last
> statistics run? Or are the previous tries counted a second time on
> the second run? Or is a complete run made the second time which
> replaces the first run?
When stats are generated they are generated for what you asked for,
you can limit them to a time wndow if you want, or they coan cover the
entire period of interaction with them.
I think though you are asking about the storage of this information
into the dynamic metadata of a problem, in that case the description
'a complete run made the second time which replaces the first run'
is an accurate depiction of what occurs.
We have discussed writing a script to do this in a batch process
internal to the system so that the problem metadata gets updated more
regularly with this info, but haven't done that yet.