Hi,

Is there any way to Sort All RTK1 By Number of Fails, descending?
This would be useful for focusing re-study for difficult Kanji.

Thanks in advance.

On the Review tab page, there's a link called "detailed list". On the detailed list, click on "failed" column header twice.

http://kanji.koohii.com/status/full-kanj...ilurecount

Excellent. Just what I was looking for.
Thank you.

It would be interesting to be able to sort on *everyone*'s aggregated fails - let's see which kanji really are difficult!

bobince Wrote:It would be interesting to be able to sort on *everyone*'s aggregated fails - let's see which kanji really are difficult!

We could just start a post for most failed kanji.

Right now I have a tie between: 切, 帝, 豪, 携, 覧.

覧 is now one of the easiest ones for me, but I initially had a lot of trouble with the positioning of the different primitives.

That's a nice idea ! My most failed kanji are :

851 臓 entrails 4 (I keep mistaking it for intestines)
864 勧 persuade 4
611 憩 recess 3
628 愉 pleasure 3
787 訟 sue 3
823 恥 shame 3
929 数 number 3
1087 礼 salute/salutation 3

Not counting the ones I fail because I get the keywords mixed up, I've had a lot of trouble with these: 鯵 楢 襖 薩 彙

[ファブリス]

bobince Wrote:It would be interesting to be able to sort on *everyone*'s aggregated fails - let's see which kanji really are difficult!

Here is the 50 most failed kanji accross all users. Kanji,framenum,keyword,SUM pass,SUM fail, ratio pass/fail

嘱 1954 entrust 1917 1044 1,8362
憾 631 remorse 6082 3307 1,8391
薦 2000 recommend 1867 1007 1,854
顕 1789 appear 2137 1142 1,8713
耗 1914 decrease 1973 1037 1,9026
唆 766 tempt 5208 2726 1,9105
褒 998 praise 4212 2191 1,9224
撤 760 remove 5138 2649 1,9396
粛 1733 solemn 2344 1194 1,9631
随 1298 follow 3314 1676 1,9773
勤 1577 diligence 2481 1228 2,0204
恭 1803 respect 2170 1063 2,0414
微 889 delicate 4759 2313 2,0575
越 387 surpass 7502 3614 2,0758
泰 1570 peaceful 2558 1216 2,1036
叙 1660 confer 2396 1136 2,1092
誘 916 entice 4542 2129 2,1334
懸 1394 suspend 2944 1376 2,1395
慶 2001 jubilation 1918 894 2,1454
塑 1960 model 1999 928 2,1541
縫 1563 sew 2659 1227 2,1671
徳 885 benevolence 4726 2164 2,1839
峰 1562 summit 2685 1226 2,19
堕 1314 degenerate 3099 1412 2,1948
託 1917 consign 1909 863 2,2121
収 1510 income 2820 1272 2,217
献 1615 offering 2405 1083 2,2207
勧 864 persuade 4552 2048 2,2227
既 1481 previously 2741 1225 2,2376
傑 999 greatness 4246 1889 2,2478
惜 1187 pity 3576 1589 2,2505
釈 1908 explanation/interpretation 2101 933 2,2519
庸 1182 commonplace/comfortable 3478 1542 2,2555
愁 901 distress 4758 2109 2,256
郭 1842 enclosure 2141 946 2,2632
謹 1576 discreet 2487 1098 2,265
逓 1858 parcel post 2108 927 2,274
敷 1883 spread 2073 910 2,278
挿 1116 insert 3861 1693 2,2806
邸 1841 residence 2188 957 2,2863
得 876 gain 4620 2019 2,2883
励 866 encourage 4773 2067 2,3091
擁 1387 hug 3007 1301 2,3113
修 1722 discipline 2316 1002 2,3114
穏 1147 calm 3688 1592 2,3166
窮 1326 hard up 3065 1321 2,3202
統 1347 overall 3077 1322 2,3275
朗 1469 melodious 2817 1210 2,3281
錯 1185 confused 3549 1521 2,3333
把 1752 grasp 2320 994 2,334

Hey, so I'm not the only one who totally hates entrust.

Actually, that is pretty interesting, because I think I've failed a good deal of those as well. Interesting commonalities.

You're nice to us.

Taking a quick look it looks like maybe "tempt" is number one. (anybody see one with more missed?) Man I know that I have contributed to that one. I wonder how many of those are because it was confused with beckon.

Currently I look like this:
1035 善 virtuous
1163 群 flock
1310 陛 highness
300 冗 superfluous
306 坑 pit
1027 匁 monme

I hate superfluous cuz I can't ever remember if its a top hat or a crown! Same with pit. I think Ive finally got virtuous on the ropes tho!

Fabrice, two questions

1. Was that table only from RTK1 list?
2. How many users have 1800+ cards in stack 5 or higher?

ファブリス Wrote:Here is the 50 most failed kanji accross all users.

Yep, I certainly have trouble with most of those.

I think a common feature that they have is that their keyword is synonymous with other keywords, making them harder to distinguish. E.g.,

把 grasp vs 握 grip
憾 remorse vs 悔 repent
耗 decrease vs 減 dwindle
越 surpass vs 超 transcend
塑 model vs 型 mould
峰 summit vs 頂 place on the head
錯 confused vs 妄 delusion

[ファブリス]

@Nukemarine

1. Yes, sorry I forgot to mention that the list was actually all RtK3 kanji (that is, the top 50 most failed). The results ABOVE are RtK1 kanji alone. Here's the results accross ALL kanji, it's all RtK3 :

槃 2522 enjoyment 97 80 1,2125
肴 2896 entreat 60 49 1,2245
赳 2936 bold 69 55 1,2545
逞 2971 tough 84 65 1,2923
剪 2941 pruning 77 59 1,3051
躊 2996 hesitate 70 53 1,3208
肇 2893 founding 69 52 1,3269
焔 2546 flames 97 73 1,3288
饅 2998 bean jam 63 47 1,3404
浩 2311 abounding 145 108 1,3426
聚 2939 crowd 95 70 1,3571
虔 2868 reserved 71 51 1,3922
鋤 2756 plow 74 53 1,3962
聘 2678 summons 101 72 1,4028
悉 2430 without exception 120 85 1,4118
蔑 2390 revile 151 106 1,4245
瞥 2927 glimpse 70 49 1,4286
瑞 2563 fortunate 142 99 1,4343
瑳 2559 burnish 103 71 1,4507
嚢 2883 cyst 93 64 1,4531
禦 2597 fend off 102 70 1,4571
黎 2935 tenebrous 79 54 1,463
祢 2601 ancestral shrine 101 69 1,4638
頓 2788 immediate 104 71 1,4648
譬 2728 parable 85 58 1,4655
鑓 2767 javelin 94 64 1,4688
錐 2760 awl 90 61 1,4754
鮒 2825 crucian 65 44 1,4773
穆 2612 obeisant 102 69 1,4783
欣 2888 elation 77 52 1,4808
鄭 2425 courtesy 126 85 1,4824
詑 2718 prevarication 79 53 1,4906
敦 2931 empathetic 79 53 1,4906
翫 2899 fiddle with 82 55 1,4909
簸 2647 winnow 88 59 1,4915
讃 2711 compliment 87 58 1,5
甦 2940 resucitate 75 50 1,5
稀 2611 sparse 125 83 1,506
樗 2526 plotosid 101 67 1,5075
丞 2919 helping hand 101 67 1,5075
鹸 2867 lye 83 55 1,5091
晋 2436 progress 121 80 1,5125
榛 2475 hazel 129 85 1,5176
畢 2570 lastly 126 83 1,5181
毅 2889 stalwart 76 50 1,52
斡 2781 auspices 76 50 1,52
葵 2391 hollyhock 140 92 1,5217
楢 2493 Japanese oak 102 67 1,5224
誼 2719 familiarity 96 63 1,5238
潰 2319 defile 154 101 1,5248

Far less users are doing RtK3. Perhaps the lower pass rate is due to spending less time on getting the stories right, or maybe some users are trying to push through without a story altogether?

PS: My thoughts based on both RtK1 and RtK3 results is that the keywords are the more difficult ones, not just in themselves but also because those are similar to previously learned keywords. The latter situation is especially true for RtK3 with already 2042 keywords previously learned.

2. Out of 2514 users who have at least one flashcard in box 5 or higher, 175 users have 1800+ cards in box 5 or higher, 480 users have 1000+ flashcards in box 5 or higher, 812 users have 500+ flashcards in box 5 or higher. It seems linear but I'm not seeing it as a graph. I'm not very good at SQL querries though. Perhaps the results would be more relevant by doing a ratio of the users total of cards in box 5 and higher over the user's total of flashcards. That would show how many users use the Leitner system fully, regardless of how far they got into RtK. I don't know how to run two COUNT's in a query already using a GROUP BY clause on the users though. If somebody knows how to write more advanced queries by all means, post them and I can try them out.

The trick is to build a subquery.

For example, say you have two queries:

SELECT a, COUNT(*)
FROM tbl
GROUP BY a;

SELECT COUNT(*)
FROM tbl;

You may combine them into one query if you wanted to see percentages:

SELECT a,
COUNT(*) /
(
SELECT COUNT(*)
FROM tbl
) AS pct
FROM tbl
GROUP BY a;

For simplicity, I have omitted WHERE statements, but you can stick those in to count only the cards that you care about. The syntax above works with MySQL.

Also, if the query is too complex to do it in one shot, you can always write a small stored procedure to do the calculation. If this isn't clear enough for you to figure it out, post your individual queries (or e-mail them to me) and I'll piece it together for you.

Good luck!

None of the most missed kanji (entrust, remorse, etc) really lend themselves to visuals. They're all primarily mental actions or feelings.

Fabrice, something tells me there's a thesis paper hidden in all that data you collected via RevTK.

Interesting results all around. I agree less visual kanji makes for harder retention. Also its interesting only 7% of the users have 1800 cards (very rough estimate to say 90% of rtk) in stack 5 or higher. I do think this points out that some users do RTK and RevTK only up until they "finish" book 1 (put in all stories).

I'm not a statician though. So I won't be looking to deep into this.

Samsara Wrote:None of the most missed kanji (entrust, remorse, etc) really lend themselves to visuals. They're all primarily mental actions or feelings.

I have noticed this, too, when going from the sub-1000's to the second half: there appear to be fewer abstract/hard words in the 1000-2000 range. Correct me, RTK3 veterans, if I am wrong, but there seem to be even fewer fuzzy-words in the 2000-3000 range, but many more esoteric, but concrete keywords.

Nukemarine Wrote:I do think this points out that some users do RTK and RevTK only up until they "finish" book 1 (put in all stories).

Perhaps people tend to switch to an external srs when they start working on readings/vocabulary/sentences, to keep everything in the same place? That's what I did, but I don't know exactly how common it is.

Thanks Fabrice! Groan - I recognise a lot of that list from my own perennial failures!

Surprised not to see more of the types-of-tree though, I can never remember which is which!