How to use the weak unit trend checker
After a few tests, this page sorts out which units keep holding you back. Put a unit name and its error rate for each sitting on one line, set a threshold, and the tool separates units that cross it again and again from those that slipped only once. Unlike ranking the wrong answers from a single test, it looks at how each unit moved between sittings.
For every unit it computes the average error rate, the latest sitting and the change from the first sitting to the latest. It counts sittings at or above your threshold; reaching your recurring count makes the unit a recurring weakness, one or more makes it a one-off slip, and none at all makes it steady. The threshold includes its boundary, so with a threshold of 30% a sitting at exactly 30% is counted. The table lists recurring weaknesses first, then sorts by average error rate.
Units may have different numbers of sittings; each is averaged over its own line, and mixed counts are flagged. Because only the rate is used, a unit with few questions swings a long way on a single item. The page does not schedule reviews and does not forecast the next sitting's error rate.
Frequently asked questions
By how many sittings landed at or above your error-rate threshold. Reaching the recurring-count you set makes it a recurring weakness; one or more but fewer than that is a one-off slip; none at all is steady.
Yes. Each unit is averaged over the sittings on its own line. When the counts differ, the page points that out separately.
It counts as at or above. With a threshold of 30%, a sitting at exactly 30% is counted as crossing it.