"You're always getting this wrong, take note"


#1

Hi,

Sorry if this has been requested or exists - couldn’t find something like it in the list.
There are some items that have been in my review pile for /months/ … I mean, months. They never seem to progress above Master level. 
Can someone come up with a script that recognises a review that you just got wrong (yet again) as being something that was unlocked AGES ago and alert you for you to note it in a list of your own.
I found that if I can collect these items in to a list I start to see why I mix them up and get them wrong, and find ways to sort it out, I get out of the rut, but if they are spaced too far apart I can’t figure it out.
Therefore it would be handy to have something pop up that says “You went back to apprentice on this, and it’s over 100 days old” or some such, so I can take action on it.


#2

An easy catch might by longest streak being 3 for example


#3

I’d like this as well, I have more than a few items that seem to keep coming back, I usually mess them up just before burning …


#4

Yeah, I agree. There are definitely many kanji with on’yomi readings that are very similar. It can be very frustrating to keep getting them mixed up. Over time though, they start to become more and more natural. I guess, just trust in the SRS, and eventually you will always know that for this particular word, you use this reading. But yeah, I have the same problem. 

An example might be: 
= しがつ
曜日 =げつようび

That げつ vs. がつ keeps getting me on a few words.

TL:DR:
Trust that the SRS will sort it out eventually.


#5

Not sure if this is helpful, but this allows you to mark things manually so it is easier to review them after. 
/t/Userscript-Item-Marker/12914/2
Also, this shows your critical items.
/t/Shinkoku-Single-page-site-to-view-your-critical-items-at-a-glance/9356/1

there is also a thread about exporting critical items so you can import them into anki or houhou.
Can’t think of anything else atm, but these might help until someone comes up with a script that suits more to what you’re after :smiley:


#6

To be honest, it’s more the meanings that I get wrong these days rather than the readings, but the point is, I’d like a script to alert me when I go back to apprentice for older items, because they are building up, and ultimately, if I have 100 items that are 100 days old, then I have too many reviews to do in any given day.


#7
Amenophyckxs said... Not sure if this is helpful, but this allows you to mark things manually so it is easier to review them after. 
/t/Userscript-Item-Marker/12914/2

 This seems to be close to what I need hmm, thanks! :D

#8
madrory said... To be honest, it's more the meanings that I get wrong these days rather than the readings, but the point is, I'd like a script to alert me when I go back to apprentice for older items, because they are building up, and ultimately, if I have 100 items that are 100 days old, then I have too many reviews to do in any given day.

 Hmm, yeah, I guess I misinterpreted you in the first post.
That would be nice indeed, some way to differentiate between new ones that don't really stick and old ones that are slowly being forgotten.

I use a reorder script that shows me which levels my reviews are and I sort them by lowest level first, that might be a workaround for now? That way you know that the ones that are from 5 levels ago are probably the ones that are being forgotten, rather than a new one that doesn't stick?

#9

I think the script could do something like this:  If an item were over a certain number of total reviews and in the Mastered bucket, flag these for further optional review.  Interestingly, these don’t trigger the threshold for critical items because the overall success rate is sufficiently high to not flag them.  Or, alternately, look for combined percentage mastered items at a lower threshold than the critical items threshold.

I took some known difficult items for me and was rather embarrassed about how high the total number of reviews climbed.  Clearly these aren’t sticking well and somewhere in the high 30s these had started to become an issue for me because the total number of difficult to learn items meant I was spending a significant chunk of time trying to learn what I had not learned well enough to begin with.  One of these days I’m going to sort out the difference between 裁判 and 審判.  It seems pretty easy to mix up judge-judge and judge-judge, ばん - ぱん… but for a frickin’ year still mixing them up?  But I did just get 審判 correct, so やった!


#10
lsh3rd said...<the stuff above>
Yeah, the %age right isn't a good indicator really, most of my problem items are above the 74% threshhold, but the fact is, they're justified, and they're ancient (Sorry, couldn't resist using KLF lyrics), they're all from 無無 land !
So a script that says "Er, you got this wrong last week and it's now 2 months old" would be just dandy!

#11

I would like this too! Another way might be if either reading or meaning were at 15 total reviews, as that is close to having gone twice through the whole system (regardless of the % correct).

It is Apprentice 1,2,3,4, Guru 5, 6, Master 7, Enlightened 8, BURNED (9). (Correct?)


#12
AnimeCanuck said... I would like this too! Another way might be if either reading or meaning were at 15 total reviews, as that is close to having gone twice through the whole system (regardless of the % correct).

It is Apprentice 1,2,3,4, Guru 5, 6, Master 7, Enlightened 8, BURNED (9). (Correct?)

Yes, that’s the right srs_numerics (EDIT, I had said maths is needed but I see total_right on either side could be used)

 The perfect candidate for this is this Vocab which I’ve fielded via the API:
  • 感想 / かんそう / impressions, one’s impressions, thoughts
  • "srs_numeric":6,
  • “unlocked_date”:1448809125, (Sun, 29 Nov 2015 14:58:45)
  • “available_date”:1462537800, (Fri, 06 May 2016 12:30:00)
  • “meaning_correct”:28, “meaning_incorrect”:10, “meaning_max_streak”:5, “meaning_current_streak”:2,
  • “reading_correct”:28, “reading_incorrect”:0, “reading_max_streak”:28, “reading_current_streak”:28,

As you can see, I’ve never gotten the reading wrong, but the best in a row for meaning I ever reached was five.
Current streak is 2, so if I get it wrong, I go back to srs_numeric 4. Current streak < 4 and age = huge, so call the “YOU KEEP GETTING IT WRONG”.

EDIT:
Or indeed, have it (You have total_right super_high and yet you still got it wrong, alart!)


#13
madrory said...
AnimeCanuck said... I would like this too! Another way might be if either reading or meaning were at 15 total reviews, as that is close to having gone twice through the whole system (regardless of the % correct).

It is Apprentice 1,2,3,4, Guru 5, 6, Master 7, Enlightened 8, BURNED (9). (Correct?)

Yes, that’s the right srs_numerics (EDIT, I had said maths is needed but I see total_right on either side could be used)

 The perfect candidate for this is this Vocab which I’ve fielded via the API:
  • 感想 / かんそう / impressions, one's impressions, thoughts
  • "srs_numeric":6,
  • "unlocked_date":1448809125, (Sun, 29 Nov 2015 14:58:45)
  • "available_date":1462537800, (Fri, 06 May 2016 12:30:00)
  • "meaning_correct":28, "meaning_incorrect":10, "meaning_max_streak":5, "meaning_current_streak":2,
  • "reading_correct":28, "reading_incorrect":0, "reading_max_streak":28, "reading_current_streak":28,
As you can see, I've never gotten the reading wrong, but the best in a row for meaning I ever reached was five. Current streak is 2, so if I get it wrong, I go back to srs_numeric 4. Current streak < 4 and age = huge, so call the "YOU KEEP GETTING IT WRONG".

EDIT:
Or indeed, have it (You have total_right super_high and yet you still got it wrong, alart!)

 Wow! I have no idea how to look at my own API like that, but seeing your data here, that’s really interesting! (I’m no programmer. I just work with one every once in a while, and bug test the odd thing.) Does this post mean you’ve come up with a way to write a script for this?  Are you working on one? excited and hopeful :slight_smile:

 Yeah, and November to May is quite a long time. I’m sure I have a few items in that camp myself.

#14
AnimeCanuck said...
madrory said...
AnimeCanuck said... I would like this too! Another way might be if either reading or meaning were at 15 total reviews, as that is close to having gone twice through the whole system (regardless of the % correct).

It is Apprentice 1,2,3,4, Guru 5, 6, Master 7, Enlightened 8, BURNED (9). (Correct?)

Yes, that’s the right srs_numerics (EDIT, I had said maths is needed but I see total_right on either side could be used)

 The perfect candidate for this is this Vocab which I’ve fielded via the API:
  • 感想 / かんそう / impressions, one's impressions, thoughts
  • "srs_numeric":6,
  • "unlocked_date":1448809125, (Sun, 29 Nov 2015 14:58:45)
  • "available_date":1462537800, (Fri, 06 May 2016 12:30:00)
  • "meaning_correct":28, "meaning_incorrect":10, "meaning_max_streak":5, "meaning_current_streak":2,
  • "reading_correct":28, "reading_incorrect":0, "reading_max_streak":28, "reading_current_streak":28,
As you can see, I've never gotten the reading wrong, but the best in a row for meaning I ever reached was five. Current streak is 2, so if I get it wrong, I go back to srs_numeric 4. Current streak < 4 and age = huge, so call the "YOU KEEP GETTING IT WRONG".

EDIT:
Or indeed, have it (You have total_right super_high and yet you still got it wrong, alart!)

 Wow! I have no idea how to look at my own API like that, but seeing your data here, that’s really interesting! (I’m no programmer. I just work with one every once in a while, and bug test the odd thing.) Does this post mean you’ve come up with a way to write a script for this?  Are you working on one? excited and hopeful :slight_smile:

 Yeah, and November to May is quite a long time. I’m sure I have a few items in that camp myself.

 You don’t have to be a programmer, take a look at the API docs, you can just copy/paste the example link in the description into your browser, and you’ll see the data. There are many sites and plugins that can pretty format json, like this site, or this chrome plugin.

#15
tomboy said...
AnimeCanuck said...
madrory said...
AnimeCanuck said... I would like this too! Another way might be if either reading or meaning were at 15 total reviews, as that is close to having gone twice through the whole system (regardless of the % correct).

It is Apprentice 1,2,3,4, Guru 5, 6, Master 7, Enlightened 8, BURNED (9). (Correct?)

Yes, that’s the right srs_numerics (EDIT, I had said maths is needed but I see total_right on either side could be used)

 The perfect candidate for this is this Vocab which I’ve fielded via the API:
  • 感想 / かんそう / impressions, one's impressions, thoughts
  • "srs_numeric":6,
  • "unlocked_date":1448809125, (Sun, 29 Nov 2015 14:58:45)
  • "available_date":1462537800, (Fri, 06 May 2016 12:30:00)
  • "meaning_correct":28, "meaning_incorrect":10, "meaning_max_streak":5, "meaning_current_streak":2,
  • "reading_correct":28, "reading_incorrect":0, "reading_max_streak":28, "reading_current_streak":28,
As you can see, I've never gotten the reading wrong, but the best in a row for meaning I ever reached was five. Current streak is 2, so if I get it wrong, I go back to srs_numeric 4. Current streak < 4 and age = huge, so call the "YOU KEEP GETTING IT WRONG".

EDIT:
Or indeed, have it (You have total_right super_high and yet you still got it wrong, alart!)

 Wow! I have no idea how to look at my own API like that, but seeing your data here, that’s really interesting! (I’m no programmer. I just work with one every once in a while, and bug test the odd thing.) Does this post mean you’ve come up with a way to write a script for this?  Are you working on one? excited and hopeful :slight_smile:

 Yeah, and November to May is quite a long time. I’m sure I have a few items in that camp myself.

 You don’t have to be a programmer, take a look at the API docs, you can just copy/paste the example link in the description into your browser, and you’ll see the data. There are many sites and plugins that can pretty format json, like this site, or this chrome plugin.
 Thanks, tomboy! That’s handy to know. But which example link, where? (Within the API docs link you provided? Or am I missing something in madrory’s post above?) Is Json what WK is done in? (I know when Twitter changed their API, it caused a huge problem for TweetDeck. Is there more than one language APIs are written in? I always thought “changed the API” was just simply reprograming (new code)… in the same language… but maybe Twitter screwed people over that badly by switching languages!

#16
AnimeCanuck said...
tomboy said...
AnimeCanuck said...
madrory said...
AnimeCanuck said... I would like this too! Another way might be if either reading or meaning were at 15 total reviews, as that is close to having gone twice through the whole system (regardless of the % correct).

It is Apprentice 1,2,3,4, Guru 5, 6, Master 7, Enlightened 8, BURNED (9). (Correct?)

Yes, that’s the right srs_numerics (EDIT, I had said maths is needed but I see total_right on either side could be used)

 The perfect candidate for this is this Vocab which I’ve fielded via the API:
  • 感想 / かんそう / impressions, one's impressions, thoughts
  • "srs_numeric":6,
  • "unlocked_date":1448809125, (Sun, 29 Nov 2015 14:58:45)
  • "available_date":1462537800, (Fri, 06 May 2016 12:30:00)
  • "meaning_correct":28, "meaning_incorrect":10, "meaning_max_streak":5, "meaning_current_streak":2,
  • "reading_correct":28, "reading_incorrect":0, "reading_max_streak":28, "reading_current_streak":28,
As you can see, I've never gotten the reading wrong, but the best in a row for meaning I ever reached was five. Current streak is 2, so if I get it wrong, I go back to srs_numeric 4. Current streak < 4 and age = huge, so call the "YOU KEEP GETTING IT WRONG".

EDIT:
Or indeed, have it (You have total_right super_high and yet you still got it wrong, alart!)

 Wow! I have no idea how to look at my own API like that, but seeing your data here, that’s really interesting! (I’m no programmer. I just work with one every once in a while, and bug test the odd thing.) Does this post mean you’ve come up with a way to write a script for this?  Are you working on one? excited and hopeful :slight_smile:

 Yeah, and November to May is quite a long time. I’m sure I have a few items in that camp myself.

 You don’t have to be a programmer, take a look at the API docs, you can just copy/paste the example link in the description into your browser, and you’ll see the data. There are many sites and plugins that can pretty format json, like this site, or this chrome plugin.
 Thanks, tomboy! That’s handy to know. But which example link, where? (Within the API docs link you provided? Or am I missing something in madrory’s post above?) Is Json what WK is done in? (I know when Twitter changed their API, it caused a huge problem for TweetDeck. Is there more than one language APIs are written in? I always thought “changed the API” was just simply reprograming (new code)… in the same language… but maybe Twitter screwed people over that badly by switching languages!

 I highlighted the link with a red box:


You can copy and paste that to the browser and you’ll get a bunch of text back. 
JSON is the format that returned text is in, that contains the data you’re looking for. On the right side there’s an example how it looks nicely formatted.

The 1 after the last / in the link signifies the level of items you’re looking for, you can ask for more changing it to a comma separated list, 
1,2,3,4,5,6,7,8,9,10  for example in your case if you want to get all your vocab data.

#17
tomboy said...
AnimeCanuck said...
tomboy said...  You don't have to be a programmer, take a look at the API docs, you can just copy/paste the example link in the description into your browser, and you'll see the data. There are many sites and plugins that can pretty format json, like this site, or this chrome plugin.
 Thanks, tomboy! That's handy to know. But which example link, where? (Within the API docs link you provided? Or am I missing something in madrory's post above?) Is Json what WK is done in? (I know when Twitter changed their API, it caused a huge problem for TweetDeck. Is there more than one language APIs are written in? I always thought "changed the API" was just simply reprograming (new code)... in the same language... but maybe Twitter screwed people over that badly by switching languages!
 I highlighted the link with a red box:
[...helpful image removed for space...]

You can copy and paste that to the browser and you'll get a bunch of text back. 
JSON is the format that returned text is in, that contains the data you're looking for. On the right side there's an example how it looks nicely formatted.

The 1 after the last / in the link signifies the level of items you're looking for, you can ask for more changing it to a comma separated list, 
1,2,3,4,5,6,7,8,9,10  for example in your case if you want to get all your vocab data.
 Thanks! You're awesome! I'll definitely be checking that out after my next large batch of reviews! ^_^

#18
AnimeCanuck said...
tomboy said...
AnimeCanuck said...
tomboy said...  You don't have to be a programmer, take a look at the API docs, you can just copy/paste the example link in the description into your browser, and you'll see the data. There are many sites and plugins that can pretty format json, like this site, or this chrome plugin.
 Thanks, tomboy! That's handy to know. But which example link, where? (Within the API docs link you provided? Or am I missing something in madrory's post above?) Is Json what WK is done in? (I know when Twitter changed their API, it caused a huge problem for TweetDeck. Is there more than one language APIs are written in? I always thought "changed the API" was just simply reprograming (new code)... in the same language... but maybe Twitter screwed people over that badly by switching languages!
 I highlighted the link with a red box:
[...helpful image removed for space...]

You can copy and paste that to the browser and you'll get a bunch of text back. 
JSON is the format that returned text is in, that contains the data you're looking for. On the right side there's an example how it looks nicely formatted.

The 1 after the last / in the link signifies the level of items you're looking for, you can ask for more changing it to a comma separated list, 
1,2,3,4,5,6,7,8,9,10  for example in your case if you want to get all your vocab data.
 Thanks! You're awesome! I'll definitely be checking that out after my next large batch of reviews! ^_^
Let me know if I can help with anything :)
(You can find me on slack if you think that would be easier.)

#19

Hey! so check out rfindley’s stat site. There was an update that might make that easy for you. https://www.idigtech.com/wanikani/