Let's make the True Retention table look better

[Expertium]

Ah, ok

[Expertium]

@sorata

Question: Does the “True Retention” graph includes intraday learning cards?

It says “Pass rate of cards with an interval ≥ 1 day.”, so no.

[sorata]

I meant to ask the other one actually

[Expertium]

This one?

I think so

[A_Blokee]

@sorata

If you have “all history” selected on top, you’ll need to show an extra radio button like the other graphs in this screen. (We need space!!)

We may have to pick a sacrifice…
Either that or have the buttons double in rows if it’s too long which I suppose I prefer.

@sorata

• Do we need stats for “Today”? There’s overlap with the first section of the this screen so maybe we can either remove it or put some extra info there instead?

This doesn’t differentiate between learning/filtered/young cards. Maybe I could just add that info and remove it from the other graph?

@sorata

Question: Does the “True Retention” graph includes interday learning cards?

fn interval_bucket(review: &RevlogEntry) -> Option<IntervalBucket> {
    match review.review_kind {
        RevlogReviewKind::Learning | RevlogReviewKind::Relearning | RevlogReviewKind::Filtered => {
            Some(IntervalBucket::Learning)
        }
        RevlogReviewKind::Review => Some(if review.last_interval < 21 {
            IntervalBucket::Young
        } else {
            IntervalBucket::Mature
        }),
        RevlogReviewKind::Manual | RevlogReviewKind::Rescheduled => None,
    }
}

This is how it separates them exactly. Is it possible to have a review card with a less than 1 day interval? Even with fsrs5 I feel like they’re labeled learning if they’re sub 1 day.

[Expertium]

Yep, they’re labeled as learning with FSRS-5.
Your first image looks pretty good, btw. Although, the amount of of radio buttons is a bit too big.
Also, regarding removing “True” from “True Retention”, since it was called that in the add-on, and since it’s called that in Anki 24.11 already, let’s keep it.
Right now the best option that I see is removing the table, renaming Answer Buttons to True Retention, and keeping it as on your first image; with the “Pass rate of cards with an interval ≥ 1 day” message. So no UnTrue Retention, lol.

[rossgb]

I’m not sure about this.

1. It’s starting to look a bit busy with that many radio buttons.
2. Why would only “Answer Buttons” have so many selectable periods?
   (All others with periods are 1 month/3 months/1 year/[all])
3. I feel it does not really replace what “True Retention” give us.

The main thing I use “True Retention” for is seeing how close I am to my desired retention (90%) over time. Hence my suggestion of the Retention Percentage only version of the table (ignore the yucky dark green):

@rossgb

This is much more awkward to do with the “Answer Buttons” graph as you have to flip between the radio buttons to get the different retention numbers.

N.B. I do not think a graph a la “Retenion Graph” fully replaces this use either. It is also nice, but I feel it is giving me something different.

[Danika_Dakika]

@Expertium

any ideas how we can merge Answer Buttons and True Retention into the same thing?

No – but uhhh … why would we? We’ve got a major case of the scope-creeps here.

The project brief is – create a TR table that is as attractive, engaging, and easy to understand as the Helper add-on one, but in a smaller/scalable form – a pocket-size pretty version. In my view, we shouldn’t be talking about the Answer Buttons graph at all. [Whether we want to continue having both of them, whether Answer Buttons needs some updating, whether you can murder a graph with too many radio buttons – those are questions for another day, and we shouldn’t let them derail this.]

When I look at the current TR table, there are only so many things that a mind can easily capture, and there’s some noise I have to filter out. After I skip over all of the pass/fail and learned/relearned numbers, I’m really only taking in this:

@rossgb

Kudos to @rossgb for proposing that (and then re-proposing it)! That’s got my vote – with 2 changes (pretty please?):

1. I agree with Expertium that there needs to be a total to know how to properly weigh these against each other. The time periods are meaningful, but when my eyes glaze-over at all the raw numbers in the add-on table, I do still take in that they are 10s, 100s, 1000s. @rossgb , I think there’s room for one more column, isn’t there? Maybe split the column under the Total header, and give us a total count to go with each row?

2. As mentioned all around, the dark green is a problem. I don’t know what colors the other add-on table was using, or if it was as good on light mode as on dark, but those might be candidates. We should mimic the colors already being used for young and mature in main Stats, if possible. In Card counts (if I found the right thing ) – young is rgb(116, 196, 118), mature is rgb(49, 163, 84) – I think those are the same on light and dark. Those must be the same colors used at some point in the Reviews graph gradient (and in the hover). It’s not text, but it’s a good starting point. What would that look like?
   

   

Nice to have: Will it behave like the rest of the page and add a row for “All history” when that’s changed for the page?

[sorata]

@Expertium

This one?

I was asking about interday cards and whether they are in TR table. I guess they are just considered to be in review state so probably. Initially thought it was in some kind of gray zone.

@A_Blokee

This doesn’t differentiate between learning/filtered/young cards. Maybe I could just add that info and remove it from the other graph?

I think that would be better. Also, from what I know the TR add-on also showed it’s data under “Today”.

[sorata]

@rossgb

The main thing I use “True Retention” for is seeing how close I am to my desired retention (90%) over time. Hence my suggestion of the Retention Percentage only version of the table (ignore the yucky dark green):

That’s a fine idea but if you have duplicated info it’s a bit confusing. So, I think it’s best to do it in a way that doesn’t leave us with same info in multiple places.

@rossgb

This is much more awkward to do with the “Answer Buttons” graph as you have to flip between the radio buttons to get the different retention numbers.

I think it’s less overwhelming with not many raw numbers showing up together. Or the other option is, have the radio buttons select between “young”, “mature”, “total” etc. and show data for different time periods together. Clearly, the current is a compromise.

[rossgb]

Just for you:

[sorata]

@Expertium

Right now the best option that I see is removing the table, renaming Answer Buttons to True Retention, and keeping it as on your first image

Another idea: Remove the columns for Hard and Easy button and put the raw data in the pop-up. It frees up space and we can get stats for “Total” too.

I’m talking of this pop-up:

[rossgb]

@sorata

That’s a fine idea but if you have duplicated info it’s a bit confusing. So, I think it’s best to do it in a way that doesn’t leave us with same info in multiple places.

I think it’s perfectly fine to show the same data multiple times, but in different ways, when you want to show different things.

@sorata

Or the other option is, have the radio buttons select between “young”, “mature”, “total” etc. and show data for different time periods together. Clearly, the current is a compromise.

@sorata

Another idea: Remove the columns for Hard and Easy button and put the raw data in the pop-up. It frees up space and we can get stats for “Total” too.

I think the “Answer Buttons” graph is already doing a good job as it is. I wouldn’t want to change it in these ways. I would prefer a new graph to editing “Answer Buttons” like this.

I quite like this one, even though it does not show data for as many periods:

@sorata

[Expertium]

Looks good!

[A_Blokee]

I’d be happy with this.

[rossgb]

Another experiment for your consideration:

Uncluttered percentages for people like me and @Danika_Dakika :

But we still give access to all absolute values for people like @Expertium :

[Expertium]

Yep, looks good. I’m fine with this being the final version.
I like it more than this: Let's make the True Retention table look better - #31 by rossgb

[rossgb]

Maybe one final twiddle:

This effectively lets you see all three tables from the current Anki + the “Percentages Summary” I proposed earlier. (defaulting to the summary page)

[Expertium]

I feel like that’s a bit unnecessary, so I vote for this: Let's make the True Retention table look better - #36 by rossgb

[rossgb]

I’ve pushed a copy to my fork if anyone wants to have a play.

N.B. Checkout pretty-true-retention-table , not the default branch.

GitHub - rbrownwsws/anki at pretty-true-retention-table