kanji koohii FORUM
Anki % for learning / young / mature cards - Printable Version

+- kanji koohii FORUM (http://forum.koohii.com)
+-- Forum: Learning Japanese (http://forum.koohii.com/forum-4.html)
+--- Forum: Remembering the Kanji (http://forum.koohii.com/forum-7.html)
+--- Thread: Anki % for learning / young / mature cards (/thread-13665.html)

Anki % for learning / young / mature cards - wannabenihonjin - 2016-03-29

I just passed the half way mark in RTK 1 a few days ago and was wondering whats the normal % for learning/young/mature cards is.

Anki Stats

Thanks

RE: Anki % - yogert909 - 2016-03-29

Are you doing recognition or production?

I just finished RTK recognition style a few weeks ago and my accuracy is 82/83/77 for learning/young/mature. But that's also with 5 learning steps, 3 lapse steps, and starting ease at 130. It could be much worse or better depending on your anki settings. My memory is pretty abysmal too, so mmv.

Holy crap, I just noticed your accuracy is close to perfect. What's your secret?

RE: Anki % - wannabenihonjin - 2016-03-29

I do recognition only.The only settings i messed with was to put review cards before new ones. I also write every review and new kanji once, but only because it helps me memorize them, I personally don't care about handwriting kanji. The way I use anki might be a bit different from others, I try and memorize my next 20 cards the day before I will see them in anki, and use anki to never forget them. I have the same anki deck on my computer and phone, I do the one on my computer at 3 pm and the one on my phone at 9 pm, and from 3 pm-3pm I am learning the next 20 kanji with little 3-5 min study sessions throughout the day.

Hopefully I explained that well enough.

RE: Anki % - Stansfield123 - 2016-03-30

(2016-03-29, 5:39 pm)wannabenihonjin Wrote: I do recognition only.The only settings i messed with was to put review cards before new ones. I also write every review and new kanji once, but only because it helps me memorize them, I personally don't care about handwriting kanji. The way I use anki might be a bit different from others, I try and memorize my next 20 cards the day before I will see them in anki, and use anki to never forget them. I have the same anki deck on my computer and phone, I do the one on my computer at 3 pm and the one on my phone at 9 pm, and from 3 pm-3pm I am learning the next 20 kanji with little 3-5 min study sessions throughout the day.

Hopefully I explained that well enough.
Makes sense, and it's the right way to use Anki for RtK. You don't want to be going into it without knowing the Kanji, they would be very hard to remember that way. Still, that's pretty high retention...the stat for mature cards is especially impressive, people usually forget a lot more Kanji than 3%, once the intervals get long. That's a very good sign, means what you're doing is really working.

The fact that you only add 20 Kanji a day helps, no doubt, but still, you must have some really good stories..or whatever you use to remember them.

P.S. there is a principle (mostly just based in math, I think) on the SuperMemo site that says the ideal retention rate for SRS should be a lot lower than yours (somewhere in the 80s)...and that reviewing with very high and very low retention rates is less efficient. But that only applies to materials where you choose the difficulty of the questions, and you can choose to learn more difficult materials when the stuff you're learning is too easy, or vice versa.

It doesn't apply to RtK. You definitely want the best possible stories you can come up with, and keep your retention rate high. You should only start worrying about that once you move on to doing sentence or vocab decks, and stop depending on mnemonics. Then, you will want to adjust your retention rate for young cards to a little below 95% (that will translate to a retention rate below 90% for mature cards...which is the goal), by choosing cards of appropriate difficulty (I tend to just delete very easy cards, and suspend very hard cards...my young card retention rate right now is 95%, and the mature retention 88%...but it's a relatively new deck...I might have to start getting rid of a lot more easy cards, to keep retention low).

RE: Anki % - yogert909 - 2016-03-30

(2016-03-30, 6:26 am)Stansfield123 Wrote: P.S. there is a principle (mostly just based in math, I think) on the SuperMemo site that says the ideal retention rate for SRS should be a lot lower than yours (somewhere in the 80s)...and that reviewing with very high and very low retention rates is less efficient. But that only applies to materials where you choose the difficulty of the questions, and you can choose to learn more difficult materials when the stuff you're learning is too easy, or vice versa.

It doesn't apply to RtK. You definitely want the best possible stories you can come up with, and keep your retention rate high. You should only start worrying about that once you move on to doing sentence or vocab decks, and stop depending on mnemonics. Then, you will want to adjust your retention rate for young cards to a little below 95% (that will translate to a retention rate below 90% for mature cards...which is the goal), by choosing cards of appropriate difficulty (I tend to just delete very easy cards, and suspend very hard cards...my young card retention rate right now is 95%, and the mature retention 88%...but it's a relatively new deck...I might have to start getting rid of a lot more easy cards, to keep retention low).

The efficiency theory on the supermemo site is not actually based on choosing appropriately difficult material(although that is important in it's own right).  Rather is is based on adjusting your settings to allow for more or less frequent reviews i.e. more reviews = better accuracy = less reviews....  Somewhere around 70-80% is the sweet spot where additional reviews from re-learning failed cards are offset by longer intervals of the cards you passed.  So it does apply to rtk, and pretty much any other material.

I agree that he/she seems to be doing something right.  However, it is possible that the stats are not reflective of the actual stats if the desktop app and the mobile app are not synced before reviewing.  If they are not synced, he is effectively getting more than twice the amount of reviews and spending over twice the amount of time than it seems from the stats.  Of course the results are still excellent regardless.  It's also worth noting that similar study patterns can be replicated without pre-study and effectively having 2 separate decks by adjusting anki's settings and syncing before studying.

RE: Anki % - wannabenihonjin - 2016-03-30

(2016-03-30, 6:26 am)Stansfield123 Wrote:
(2016-03-29, 5:39 pm)wannabenihonjin Wrote: I do recognition only.The only settings i messed with was to put review cards before new ones. I also write every review and new kanji once, but only because it helps me memorize them, I personally don't care about handwriting kanji. The way I use anki might be a bit different from others, I try and memorize my next 20 cards the day before I will see them in anki, and use anki to never forget them. I have the same anki deck on my computer and phone, I do the one on my computer at 3 pm and the one on my phone at 9 pm, and from 3 pm-3pm I am learning the next 20 kanji with little 3-5 min study sessions throughout the day.

Hopefully I explained that well enough.
Makes sense, and it's the right way to use Anki for RtK. You don't want to be going into it without knowing the Kanji, they would be very hard to remember that way. Still, that's pretty high retention...the stat for mature cards is especially impressive, people usually forget a lot more Kanji than 3%, once the intervals get long. That's a very good sign, means what you're doing is really working.

The fact that you only add 20 Kanji a day helps, no doubt, but still, you must have some really good stories..or whatever you use to remember them.

P.S. there is a principle (mostly just based in math, I think) on the SuperMemo site that says the ideal retention rate for SRS should be a lot lower than yours (somewhere in the 80s)...and that reviewing with very high and very low retention rates is less efficient. But that only applies to materials where you choose the difficulty of the questions, and you can choose to learn more difficult materials when the stuff you're learning is too easy, or vice versa.

It doesn't apply to RtK. You definitely want the best possible stories you can come up with, and keep your retention rate high. You should only start worrying about that once you move on to doing sentence or vocab decks, and stop depending on mnemonics. Then, you will want to adjust your retention rate for young cards to a little below 95% (that will translate to a retention rate below 90% for mature cards...which is the goal), by choosing cards of appropriate difficulty (I tend to just delete very easy cards, and suspend very hard cards...my young card retention rate right now is 95%, and the mature retention 88%...but it's a relatively new deck...I might have to start getting rid of a lot more easy cards, to keep retention low).
I don't use stories that much, I tried at the start, but I found that just knowing the primitives was good enough for me. If I find a kanji I find is hard for me to memorize I might look at a story and see if it helps at all. I am very scared of going into vocab/sentences because I wont be able to memorize the next 20-25 or however many im learning per day before I see them.

RE: Anki % - wannabenihonjin - 2016-03-30

(2016-03-30, 12:56 pm)yogert909 Wrote:
(2016-03-30, 6:26 am)Stansfield123 Wrote: P.S. there is a principle (mostly just based in math, I think) on the SuperMemo site that says the ideal retention rate for SRS should be a lot lower than yours (somewhere in the 80s)...and that reviewing with very high and very low retention rates is less efficient. But that only applies to materials where you choose the difficulty of the questions, and you can choose to learn more difficult materials when the stuff you're learning is too easy, or vice versa.

It doesn't apply to RtK. You definitely want the best possible stories you can come up with, and keep your retention rate high. You should only start worrying about that once you move on to doing sentence or vocab decks, and stop depending on mnemonics. Then, you will want to adjust your retention rate for young cards to a little below 95% (that will translate to a retention rate below 90% for mature cards...which is the goal), by choosing cards of appropriate difficulty (I tend to just delete very easy cards, and suspend very hard cards...my young card retention rate right now is 95%, and the mature retention 88%...but it's a relatively new deck...I might have to start getting rid of a lot more easy cards, to keep retention low).

The efficiency theory on the supermemo site is not actually based on choosing appropriately difficult material(although that is important in it's own right).  Rather is is based on adjusting your settings to allow for more or less frequent reviews i.e. more reviews = better accuracy = less reviews....  Somewhere around 70-80% is the sweet spot where additional reviews from re-learning failed cards are offset by longer intervals of the cards you passed.  So it does apply to rtk, and pretty much any other material.

I agree that he/she seems to be doing something right.  However, it is possible that the stats are not reflective of the actual stats if the desktop app and the mobile app are not synced before reviewing.  If they are not synced, he is effectively getting more than twice the amount of reviews and spending over twice the amount of time than it seems from the stats.  Of course the results are still excellent regardless.  It's also worth noting that similar study patterns can be replicated without pre-study and effectively having 2 separate decks by adjusting anki's settings and syncing before studying.

My computer and phone decks are not synced. So I am just doing the same deck twice, but my phone deck % is actually better than my compuiter deck because I do my phone deck after the computer one. It takes me 20 mins or so to finish my reviews/new cards.

RE: Anki % - Stansfield123 - 2016-03-31

(2016-03-30, 1:18 pm)wannabenihonjin Wrote: I don't use stories that much, I tried at the start, but I found that just knowing the primitives was good enough for me. If I find a kanji I find is hard for me to memorize I might look at a story and see if it helps at all. I am very scared of going into vocab/sentences because I wont be able to memorize the next 20-25 or however many im learning per day before I see them.
You shouldn't try to memorize sentences. All you need to do is make sure you understand every word in the sentence, as well as their role in the sentence structure. On difficult words, edit the card to make sure the answer explains that stuff clearly. Other than that, you just have to jump in and start reviewing.

When it comes to input in language learning, the most important thing is volume. You want to go through lots and lots of material, fast. There are several things you can do, to make it easy on yourself:

1. put both the audio and sentence in the question...don't worry, that doesn't prevent you from practicing reading the Kanji

2. use a deck which re-uses the same words, in different sentences. Nayr's Core 5000 deck does that well, but it would probably be even better if you combined that deck with another one (like Core6K); there might be some overlap, but you can always just delete cards if it gets repetitive.

3.. suspend sentences where the majority of words are unknown...ideally, you want to only have one unknown word per sentence, but it's fine to have two if it's a long sentence, with 3-4 other words that you know well; as a beginner, you should probably suspend a significant chunk of the sentences

4. go fast...listen to the sentence, and decide whether you know the answer within a couple of seconds; there's a reason why that fail button is there: use it.

RE: Anki % - Stansfield123 - 2016-03-31

wannabenihonjin
My computer and phone decks are not synced. So I am just doing the same deck twice, but my phone deck % is actually better than my compuiter deck because I do my phone deck after the computer one. It takes me 20 mins or so to finish my reviews/new cards.
Ok, that explains the good stats a little better. It's not how Anki was intended to be used, but, since it's working, I wouldn't change it at this point. You should just finish adding all 2000 Kanji first.

But, after that, you should definitely stop reviewing that way (both the RtK deck and any future decks).

RE: Anki % - wannabenihonjin - 2016-03-31

(2016-03-31, 7:03 am)Stansfield123 Wrote: wannabenihonjin
My computer and phone decks are not synced. So I am just doing the same deck twice, but my phone deck % is actually better than my compuiter deck because I do my phone deck after the computer one. It takes me 20 mins or so to finish my reviews/new cards.
Ok, that explains the good stats a little better. It's not how Anki was intended to be used, but, since it's working, I wouldn't change it at this point. You should just finish adding all 2000 Kanji first.

But, after that, you should definitely stop reviewing that way (both the RtK deck and any future decks).

I was planning on stopping after I finished RTK and just do my reviews on one device. I am curious as to why doing double the reviews is a bad thing? Seems to be working well for me. Thanks for all the help

RE: Anki % - pm215 - 2016-03-31

(2016-03-31, 12:02 pm)wannabenihonjin Wrote: I was planning on stopping after I finished RTK and just do my reviews on one device. I am curious as to why doing double the reviews is a bad thing? Seems to be working well for me. Thanks for all the help
The fundamental idea behind Anki is that reviews can take up an enormous amount of time and so you want to make as efficient use as possible of them. So it carefully schedules reviews to happen only just as often as they need to (more or less) for you to remember them. Reviewing on two devices means Anki doesn't really know how often it's showing you cards, so it can't do a good job. More generally, doing double the reviews is bad because it's making bad use of your time -- you could instead be memorising more things faster by looking at them less frequently. And given the sheer mass of vocab to be learnt it's well worth trying to do it as effectively as you can. (In fact as noted earlier having a very high correct % is suggestive of inefficiency.)

RE: Anki % - wannabenihonjin - 2016-03-31

(2016-03-31, 3:56 pm)pm215 Wrote:
(2016-03-31, 12:02 pm)wannabenihonjin Wrote: I was planning on stopping after I finished RTK and just do my reviews on one device. I am curious as to why doing double the reviews is a bad thing? Seems to be working well for me. Thanks for all the help
The fundamental idea behind Anki is that reviews can take up an enormous amount of time and so you want to make as efficient use as possible of them. So it carefully schedules reviews to happen only just as often as they need to (more or less) for you to remember them. Reviewing on two devices means Anki doesn't really know how often it's showing you cards, so it can't do a good job. More generally, doing double the reviews is bad because it's making bad use of your time -- you could instead be memorising more things faster by looking at them less frequently. And given the sheer mass of vocab to be learnt it's well worth trying to do it as effectively as you can. (In fact as noted earlier having a very high correct % is suggestive of inefficiency.)

I see.I never really have a lot of reviews ( my max so far is 110) so I guess I havent had that trouble yet. I do plan on doing only one device once I finish RTK. Thanks for the info.

RE: Anki % - Stansfield123 - 2016-04-01

(2016-03-31, 12:02 pm)wannabenihonjin Wrote: I was planning on stopping after I finished RTK and just do my reviews on one device. I am curious as to why doing double the reviews is a bad thing? Seems to be working well for me. Thanks for all the help
It's only a bad thing because it takes more time. If you don't mind that, it's not a bad thing. Especially if it's only 2000 cards, the way it is with RtK. You're gonna be done in 100 days, which isn't enough time for the daily reviews to accumulate.

However, after you finish RtK, you have to realize that, according to many people, it takes going through about 10,000 sentences (in addition to immersion) to get to a level where you actually understand everyday Japanese. Add in the 2,000 RtK cards, that's a total of 12,000. If you stick to this pace of 20 new cards/day, that would take 600 days. And, over the course of 600 days, the number of daily reviews would start to add up (even if you continue having very low fail rates, the number of daily reviews will still go up a little every day, and, by the end, little by little, it's gonna be up there...you're gonna be left with no time to do immersion, which is just as important...not to mention, there's a good chance you'll burn out).

So, it becomes a lot more important to reduce the amount of time spent on reviewing your Anki decks each day. You'll definitely want to make sure all your devices are synced, and that, like yogert explains above, you change some of the Anki settings that determine the intervals on cards.

That said, I am conflicted about how much you should change those settings. I'll just share two conflicting thoughts on this, and let you make up your own mind.

1. Mathematically, the ideal fail rate (the fail rate at which you spend the least amount of time reviewing cards, assuming you are able and willing to review cards at a consistent speed, no matter how hard or easy the questions are) is around 25% (you'll have to check the exact number on the SuperMemo site). So, as per the SuperMemo site, you should set your interval modifier and easy bonus however high it takes to achieve that fail rate.

2. I hate that theory (or rather, I hate following it, I don't hate the theory, the theory itself is mathematically correct). The reason why I hate following it is twofold: for one, I hate failing at a 25% rate, it's frustrating...I like to have a higher success rate, it makes it more enjoyable to review stuff with Anki; second, I hate having to stick to a constant pace while reviewing. Instead, what I like to do is take a little longer when a card is more difficult, to try and figure out the answer. I find it hard to force myself to fail a card if I can't figure out the answer in two seconds. I do try to speed up my reviews, but I'm not THAT strict about it. And yes, I realize neither of these things is the ideal, most efficient way to review decks, but I like it better, and I'm willing to sacrifice some efficiency for that.

So, what it comes down to, is how much efficiency you're willing to sacrifice. What I do, is leave the interval modifier unchanged (at 100%, which is the default setting in Anki...that means that when you press "3" on your review, your interval doubles). The only setting I change is the easy bonus...that one, I like to set pretty high...right now, I have it at 150%, but it's been as high as 200% in the past.

In conclusion, you definitely want to make your reviewing a little more efficient than it is right now, after you finish RtK, but by how much is up to you. Just be aware that, when your fail rate is lower than 25%, you are sacrificing time in exchange for the psychological benefits of a higher success rate and a comfortable pace.

RE: Anki % for learning / young / mature cards - cophnia61 - 2016-04-01

If, after your daily review of due cards, you want to do more reviews, anki gives you the possibily of reviewing with custom decks, based on different kinds of options. Like cards failed most frequently or things like that. Or, you can take a japanese website, book, anime subtitles, and use it to review kanji. I don't know if what you do is detrimental per se, but I think you can spend the same time in a better way! (just my 5 cents Tongue )