Inconsistent stats website vs mobile + bug in graph

Hi

My quick context: last month I started using Skritter again after years of inactivity (and I’m loving it with the new interface).

I love progress stats, I find them motivating to look at. However I found some issues as per below. Note that I did a full log-out + log-in + re-sync on everything first to make sure all was synchronized.

1) The stats on the mobile version are inconsistent with the stats on the website version:

Mobile:

Website:

Hours studied: 90h01m50s (mobile) vs. 86h01m01 (website)

Total words: 709 (mobile) vs. 231 (website)
=> Even if the definition of “word” is different (how?) on the website, the total items learned shows as 694, which is still not the same as 709.

Unique characters: 451 (mobile) vs. 463 (website)

The only things that seems to match is the nr. of reviews (which would suggest to me that at least some of the data is fully synched).

2) The progress graph on the mobile version is not correct

The graph shows as being flat for the past years, however this is not correct since I’ve been adding (and learning) many new words and characters in the past month so the end of the graph can’t be flat.

Thank you for looking into this!
Dieter

:slight_smile: :slight_smile:

The stat calculations on the website and mobile apps for characters and words are slightly different. This will eventually be resolved when we update the website to be a more similar experience with the mobile apps.

1 Like

Thanks for the reply, I appreciate it.

What is the definition of a word on the mobile version?
Any plans to fix the graph issue?

Thank you
Dieter

A word in the mobile app is calculated based on the number of characters. If it’s it’s equal to 1 character than it’s a character if it’s 2 or more character than it’s classified as a word. We recognize that some individual characters could also be considered a word but for simplicity we’re just counting them as characters.

What’s wrong with your graph?

A word in the mobile app is calculated based on the number of characters. If it’s it’s equal to 1 character than it’s a character if it’s 2 or more character than it’s classified as a word. We recognize that some individual characters could also be considered a word but for simplicity we’re just counting them as characters.

“Total Words” on the stats page on mobile currently shows as 860:

  • I for sure don’t have 860 words of 2 or more characters, so is this then <nr. of single characters> + <nr. of words of 2 or more characters>?

  • Also, I assume this simply counts the items that have been marked as “Learned”? Or do you need to get them correct in reviews according to certain criteria before they are counted here?

What’s wrong with your graph?

See the second topic in my initial post:

2) The progress graph on the mobile version is not correct

The graph shows as being flat for the past years, however this is not correct since I’ve been adding (and learning) many new words and characters in the past month so the end of the graph can’t be flat.

To illustrate this further, below is a screenshot of today: the nr of words has increased to 860, yet the graph remains flat and exactly the same as before (see screenshot of my initial post where it is at 709):

Thanks
Dieter

Have you reset your account or reset specific decks from within the mobile app?

Have you reset your account or reset specific decks from within the mobile app?

No, I haven’t used that functionality (but it had been a really long time (years) since I had last logged in before I started again last month).

This is possibly the reason the chart appears to have gone flat. If you’re jumping back in after an extended break and have marked some things as forgotten then it’s possible old things are getting calculated as no longer learned. This might make it harder to have the chart show positive progress even when learning new things. We’re going to work on improving this as we transition the website to an experience more inline with the mobile apps. I’d say as long as your unique characters and total words value is increasing then all should be well under the hood.

This is possibly the reason the chart appears to have gone flat. If you’re jumping back in after an extended break and have marked some things as forgotten then it’s possible old things are getting calculated as no longer learned. This might make it harder to have the chart show positive progress even when learning new things. We’re going to work on improving this as we transition the website to an experience more inline with the mobile apps. I’d say as long as your unique characters and total words value is increasing then all should be well under the hood.

I have completely caught up on all my cards, have added many new ones and don’t have anything due, so this seems unlikely to me? Should the graph then also not go down instead of being flat?

What is the formula behind the graph? What is it actually showing? I thought the graph was a graphical reflection of the history of the “Total words” count? So this is not the case?

Having the same problem. Caught up with all my cards. Mobile up is showing I know 5000+ words. Web app shows stats incorrectly.



What is the formula behind the graph? What is it actually showing? I thought the graph was a graphical >reflection of the history of the “Total words” count? So this is not the case?

@josh, any comment on this?

@szovoroli can you describe your study habits a bit more? For example, do you primarily study using the website or apps? Have you recently reset your account or an individual deck from within app?

@dverlaeckt the graph indicates all words that have been learned over time. It’s based on some older logic from the legacy and current website which makes a determination when a word has been learned. It should be something like after you add/learn a new word then review it at correct at least once 24 hours after initially learning it then it gets marked as learned. We’re working on simplifying this system so that it’s more transparent what is considered a learned word versus one that is still being learned. It’s possible if you recently add/learned a bunch of new things they haven’t been reviewed enough to be considered learned in our old chart system.

The total words in the app is easier to explain because it’s just the total of words that have been added/learned and doesn’t have any fancy logic for whether or not they have been reviewed correctly over a period of time.

1 Like

I did not reset my account.
I had a long break from skritter and recently restarted using it. It looks like the new progress is not in the stats or on the chart in the web app. I changed from simplified only to both simplified and traditional characters.
I use the mobile app for reviewing words and adding them to my custom deck using the web app.

Thanks for the explanation. Could it be that the logic requires the word/character to be marked as learned for writing? Because I have stopped learning to write items (a bit weird perhaphs to use Skritter this way, but it still works well for reading / tones / pinyin and I’ll still have the option later to start writing again).

It looks like you’re correct about this in the mobile app and sorry I didn’t catch that earlier. It currently only adds up the learned/learning writing values from our legacy stats endpoint which would explain why your chart has stalled out. It also looks like it’s factoring in legacy characters and words, just not words. Here is a little code snippet:

const dayTotalLearned =
            day.legacyStats.word.rune.learned.all +
            day.legacyStats.char.rune.learned.all +
            day.legacyStats.word.rune.learning.all +
            day.legacyStats.char.rune.learning.all

@szovoroli just checked your account settings and it looks like you might fall into this category as well.

I don’t have a great solution for this right now but in the future as we transition further away from our legacy API we’ll be creating new charts that are based on more flexible data points.

Yes I fall into this category as well. I turned off studying writing too

Ok, at least we understand now what is going on. Thanks for looking into it.

It would be great to have a consistent experience across website and mobile and a clear definition of what the different stats mean. Perhaps in the future you can add a filter to the stats where the user can select reading / writing / tones / definition, so they can decide themselves what they want to see.

This topic was automatically closed 30 days after the last reply. New replies are no longer allowed.