Going through my review history (~140k reviews) I've discovered that the recall rate is significantly worse
if the last rating was "hard".
I'm currently testing a new version of the algorithm (which can be enabled with (setq org-fc-algorithm 'sm-v2))
but after 17k reviews with it, I still don't have enough data points to tell if it is better than the previous version.
If anyone is interested in helping with this, please leave a comment on this issue.
I'd need the review history file and a list of all card / positions.
If enough users are interested in helping with this,
I'll write a script that packages this data in a single file,
removing / replacing the filenames as these could be considered sensitive information.
Going through my review history (~140k reviews) I've discovered that the recall rate is significantly worse if the last rating was "hard".
I'm currently testing a new version of the algorithm (which can be enabled with
(setq org-fc-algorithm 'sm-v2)
) but after 17k reviews with it, I still don't have enough data points to tell if it is better than the previous version.If anyone is interested in helping with this, please leave a comment on this issue.
I'd need the review history file and a list of all card / positions. If enough users are interested in helping with this, I'll write a script that packages this data in a single file, removing / replacing the filenames as these could be considered sensitive information.