"I’m trying to fine-tune my learning steps in the Anki scheduler. Currently, my last step is 15 minutes. I want to determine how effective that is by analyzing how many of the cards that graduate into the young queue today are forgotten tomorrow — specifically, how often I press “Again” on their first review as young cards.
In other words:
Suppose today I study 100 new cards and go through all the learning steps (ending at 15m). These cards graduate into “young” and are scheduled for review tomorrow. Tomorrow, I review them — and maybe I forget 10% of them (i.e., press “Again” on 10 cards). That’s the metric I care about: how many cards that just passed the last learning step yesterday I forgot on their very first review.
My goal is to determine whether 15 minutes is a sufficient final learning step, or whether I should adjust it — maybe increase it to 30m, or lower it to 10m, depending on the forgetting rate.
I know that the “Again” count for young cards is shown in the Stats window, but that seems to include all young cards, not just the ones that were reviewed for the first time after graduating from learning the day before.
Is there any way to track that specific stat?" - I didn’t know how to articulate that questions properly, so I asked AI to help me with this.
Hi.
Basically, how to find out if my last learning step is too short/long? Like, do I need to increase 15m to 20m, or I can reduce it to 10m?
There is an add-on which shows learning steps retention.
But I want to know how many times did I click ‘again’ bottom for the first time on cards which I graduate yesterday. Not on all young cards. In order to learn how effective my last learning step be it 15m or more/less.
As you can see, I failed 1 young card and passed 32 young cards today.
You seem to be asking about something similar, but different though, considering this:
and this:
Maybe the following is sufficient for you?
Seeing Young card Stats only for cards introduced and studied yesterday
According to the manual (Searching - Anki Manual), you can search using the following strings:
introduced:1
cards answered for the first time today.
and
prop:rated=-1
cards answered one day ago.
E.g. consider the following search: prop:rated=-1 introduced:3. This will search for cards that I introduced three days ago and answered yesterday. You’d put the search into the search bar in the stats screen (where the radio buttons are) and can then look at the same table as outlined above:
As you can see, I only added one card in the last three days, that was also studied yesterday. You can also see that I passed this card.
It’s not that something is wrong with your question. If people don’t know the answer, then they can’t help you either. A lot of questions are asked quite frequently – yours however is quite unique. Thus, you’ll sometimes have to wait a while to get an answer.
This add-on adds statistics showing the percentage of correct answers for each learning step and review interval. Those can be used to tune learning steps and interval modifier.
Oh man, this add-on reminds me of one user who had LS starting from 1m to 50yrs because they were calculating retention at each step. I guess experimenting with multiday LS is fine though.
If you are using FSRS (which you should) – you might want to take a look at Step Stats in the FSRS Helper add-on. That will give you a summary of when you study a card after each initial grade, and how successful you are at that time point. https://ankiweb.net/shared/info/759844606
