Since real use conditions did pop last week. Here is some bugs or improprements for statistic dasboard.
On this first screenshot, one can see a course dashboard for teacher :
All integers inside the table seems good (calcul and sense)
"Tentative" (the number 51) should be "Évaluations"
"Moyenne tentative" (the number 3,64285714....) should be "Essais par apprenant" (and a %.2f on display)
"Moyenne note" (number 75) should be "Note moyenne" (and could be displayed 75 / 100)
"Moyenne durée" should be "Durée moyenne"
The calcul (I mean the final formula) of the "Durée moyenne" seems good to me but the acquisition of this working duration should be capped. This is hard to set. Old statistics for PL let me think that the giving up duration is 2 minutes in means ; but there exist exercices for which one really need long time completing them (20 minutes at maximum for advanced programming). For these reasons, I really think the good way to evaluate working duration is the sum of duration between evaluations but if a time gap exceed 20 minutes (like severals hours), thus this should be replace by 2 minutes which is the statistical giving up time before closing any learning application.
Here is the exercice from which come from the previous statistic. So this second screenshot present the exercice dashboard inside the ressources.
"Note moyenne" could be displayed like 75 / 100 (recall the note are out of 100 is not so bad)
"Durée moyenne" is not divided by the number of users having effectively try the exercice. It could be fine to divided it since this dashboard indicate the average time a student will need to pratice the exercice if the reader choose to use it with his students.
Globally, stats are fine for these first massive deployments and everythings seems to be computed with no any latency.
Since real use conditions did pop last week. Here is some bugs or improprements for statistic dasboard.
On this first screenshot, one can see a course dashboard for teacher :
The calcul (I mean the final formula) of the "Durée moyenne" seems good to me but the acquisition of this working duration should be capped. This is hard to set. Old statistics for PL let me think that the giving up duration is 2 minutes in means ; but there exist exercices for which one really need long time completing them (20 minutes at maximum for advanced programming). For these reasons, I really think the good way to evaluate working duration is the sum of duration between evaluations but if a time gap exceed 20 minutes (like severals hours), thus this should be replace by 2 minutes which is the statistical giving up time before closing any learning application.
Here is the exercice from which come from the previous statistic. So this second screenshot present the exercice dashboard inside the ressources.
Globally, stats are fine for these first massive deployments and everythings seems to be computed with no any latency.