[STATS] Statistics site

First, thank you for this website – it makes charting my progress so pleasant.

I’d like to report an issue I am seeing with Level 7 vocab, 「〜丁目」 is how it exists on WK, but on the stats tracker it is listed as 「丁目」 without the 〜, and as such appears to not notice that I have in fact learned that vocabulary item, so my level 7 vocab is stuck at “100/101”.

@platypus @Helix @jsnow,
It should be fixed now. The real url is:

https://www.idigtech.com/wanikani/
rather than
wkstats

For some reason that I can’t seem to identify, google says it is indexing my site, but the correct url won’t even show up in the dump of all indexed links. So, I’ve been trying to redirect all of the wrong links to the correct destination. In the process, I apparently wrote a rule in such a way that “wkstats” (without the trailing slash) was serving up a page rather than redirecting, and so it was looking in the wrong directory for support files.

2 Likes

Thank you. I really love the statistics site, you do great work :sunny:

1 Like

There’s a new API?!

It’s mentioned in the stickied post at the top of this subforum. The title of the thread is “API V2 Alpha Documentation 2”

@rfindley, I’m still seeing 丁目 instead of 〜丁目 on your application, I clicked your link directly and logged out/back in. In the interest of being thorough, I tried Firefox instead of Chrome to shake any possible cache out of the tubing and I have the same luck.

API Key: (using my V1 API key.)
OS: Fedora 25 (Linux 4.12.8-200.fc25.x86_64)
Browser: Version 60.0.3112.101 (Official Build) (64-bit)
rpm: google-chrome-beta.x86_64 60.0.3112.78-1 google-chrome
Browser2: firefox.x86_64 55.0.1-1.fc25

Please let me know if you would like any further information.

Thanks again!

1 Like

@jsnow,
Eh, sorry, I forgot to push that file up. It’s live now. You’ll need to clear cache or wait an hour.

If you only want to refresh that one page’s cache, instead of clearing all page cache, you can:

On Chrome: Open the Developer Tools (F12), go to Network tab, and check the “Disable Cache” checkbox, then F5 to refresh the page.

On Firefox: Open the Developer Tools (F12), click on the “Toolbox Options” on the right:
image
and under “Advanced Settings”, select ☑ Disable HTTP Cache (while toolbox is open)

2 Likes

@rfindley, All healed, thanks!

HI,

I recently reset some burned items I’d forgotten to Apprentice, which showed up on the stats site. I have since levelled those up to Guru, but this has not updated in my stats:

E.g. 切 shows as Apprentice on the Stats site:

But shows up as Guru on WK:
image

If you haven’t already, try the Force Refresh in the user menu of the stats site.

As I briefly tried and abandoned Wanikani in 2014 and am only now starting again, I’m now estimated to reach level 60 by 2214. I’ve got time. :smiley:

4 Likes

@felix330,

You can click on long levels on the “Level Duration” graph to exclude them from calculations.

1 Like

This isn’t related to that post… But recently, and most certainly yesterday I’ve noticed a huge change…

I’ve only done about 4-5 lessons from my current level (10 at most) so have done very few kanji (1??? 3?) … However my estimated level up time at the top is 6 days 20 hours - which I know is the fastest possible time if having done all lessons right out the gate. (I didn’t, I did ten, and haven’t done any more for about 30 days). What happened to the algorithm that used to predict based on your furthest kanji (of the 90%) out? Did you change it intentionally? I don’t understand the reason behind this change. It felt much more accurate before.

Or should I be plugging in V2 API now instead of V1? Is that the cause?

I’m confused.

Thanks! (I will try to read the 26 new replies before this…) Read them, and so far I seem to be the only one saying this. Weird.

@AnimeCanuck,

I haven’t changed the calculations for estimates. Your data looks correct. Here’s why:

Suppose your average is 20 days, and you’ve been sitting at your current level for 18 days without unlocking everything.
According to your average, you should level up in 2 days (18d + 2d = 20d).

But since you still have locked items, the soonest you can level up is in about 6 days. So, rather than estimating that you’ll level up in 2 more days, which isn’t possible, it says you’ll level up in 6 days, or 24d total (18d + 6d = 24d).

Make sense?

Quick and easy; that did the trick. Thanks again for an indispensable companion to WK.

But I haven’t even done all my kanji lessons - so they’re more than 3 days out from being Guru’d…

I really seem to remember it working differently, and pushing the date forward if I kept getting that last kanji wrong.

…? Was that not actually ever the case? (Or if so, was that not intentional?) Because I seem to recall it, and feel like that was both more accurate and beneficial for planning.

The time to level up is from now, not from when you started the level. I assume you haven’t done all your radicals for the level yet. Therefore, if it’s impossible to hit your average level up time, your time to level up is 6 days 20 hours. This assumes you do all your radicals now and all your kanji by the time the locked kanji are unlocked, and that you do all level-up-critical reviews right away and correctly.

@AnimeCanuck,

^^^ yes, this is true.

This is only true in certain circumstances.

I’m hoping this will help visualize what’s going on:
Think of your items as members of a track team in a group-marathon, where you win when 90% of your team members cross the finish line.

The race starts at the beginning of a WK level. But each runner doesn’t start running until you ‘unlock’ them.

Normally, we just rely on the team average (your average level-up time) to estimate when they will finish. There are a few reasons for that:

  • The team members are different in every race (it’s a new level, with new items), so the new team members don’t have a track record that we can use.
  • No matter how well the team members are doing at any instant, our average tells us that some of them are going to stumble a certain percentage of the time (i.e. wrong answers on reviews).

So, we rely on the average most of the time. If someone stumbles, we just assume that the average accounts for that.

But what happens if the team is taking longer than normal? By definition, the average is no longer a good estimate, so we have to evaluate things differently.

Since the team wins when 90% of the members have crossed the finish line, which guy is the most important one to look at in the race? It’s the guy that’s 90% of the way toward the back. If the guy in front stumbles, it doesn’t matter, because he’ll pick himself up again long before the guys at the back can catch up – i.e. you’re only as fast as your slowest member. The guys in the front are Apprentice4 kanji, the guys in the back are locked radicals. In fact, locked radicals haven’t even left the starting line yet.

So, depending on where your team members are in the race, a stumble may or may not affect the estimated time.

The reason your estimate is currently sitting at 6d20h is:

  • Your current level is taking longer than average, so we switch calculation methods, and…
  • …6d20h (plus the time already passed) is the fastest that the radicals at the starting line can finish. As long as more than 10% of your radicals are locked, it will sit at 6d20h.

Once 90% of your radicals are unlocked, then the 6d20h will start counting down. If you answer wrong on a low-SRS item (e.g. Apprentice1 radical), the estimated completion will increase, though never higher than 6d20h. If you answer wrong on a higher-SRS item (e.g. Apprentice4 kanji), it probably won’t affect the estimated completion time, unless all of your items are at about the same SRS level.

That makes me wonder… because I thought I only did 1-3 radicals… how few radicals are there for level 16? O.o

Because I really barely did anything.

… See, I was expecting that “last guy who’s supposed to be 90%” to be holding me back… so… that’s where the whole question arose from. I’m going to check the level right now.

… And the answer is: There are only 6, and I have done 2 of them. Perhaps 90% of the kanji were unlocked with the level. Umm… I don’t know.

But… I’ve literally only done 1 kanji - and it’s not Guru’d because I got it wrong yesterday.

So… is 16 a shorter-than-usual level and that’s messing with my sense of how the Stats site works?

Lvl16 is a normal level (6d20h). Normal levels used to be 7d20h, but that changed maybe 6 months ago.

Without excluding any levels, your avg is currently 47d13h.
Current level is 41d2h.
Since you have locked radicals, the earliest you can finish Lvl16 is (41d2h + 6d20h) = 47d22h (total), which is higher than your avg. So, the avg isn’t being used, and ‘fastest’ is being used instead (because ‘fastest’ gets you closest to your avg).

So, the slowest runner is a locked radical, which is 6d20h, assuming you do reviews right now. Any reviews you get wrong right now won’t matter… not until more radicals are unlocked.