What is the level distribution on here?

It’s a bit of a problem with bar segments on logarithmic scale. The segment at the bottom looks a lot larger than the one at the top even if the one at the top might actually have a larger number. I included the logarithmic scales mostly since they make it clearer that the drop off in users per level is roughly exponential (it’s a line in the logarithmic graph), but that does mean that reading the exact values for the sub-segments becomes a bit harder :grin:

7 Likes

Thanks for explaining! I don’t have any likes, but plz take this!

1615915784_1650928_gif-url

4 Likes

Ooh log scales! Looks like there’s a slight bump (or reduction of the reduction anyway) at 10, 20, 30, 40, 50. Slight round-number goal effect?

I’m sure this was mentioned before, but I’m too lazy to scroll back: “free” users at high levels? Are we assuming that means they quit for good at that level and let their subscription run out?

3 Likes

The data is based of the forum’s categories, which divide users based on subscription type. You’re marked as free if you don’t have any subscription active, so free users above level 3 must have let their subscription run out. I assume the majority of them quit, although there might be a small percentage who later resubscribes.

Since I still have a dataset from November 2020 I could quickly check how many of the free users above level 3 resubscribed in that time if you want to. It could give us a rough idea of how many users end up resubscribing later on.

9 Likes

Great job. I enjoy these quite a bit, feels good to beat the odds.

4 Likes

I just think, well it depends on what we’re trying to see I guess. “How many people have ever achieved” level X would be the bar that shows on these graphs, PLUS all the ones above it. “How many people are actively on” level X seems like it would be the bar on these graphs MINUS the “free” users, because they’ve quit already.

Not a criticism, just, you know, thinking to myself.

some fraction of the “lifetime” subscribers probably aren’t actively working on it either, but there’s no way to know

3 Likes

For the active users I could make a second graph, where I simply remove all users who didn’t level up since the last time I fetched the data a little over a year ago. It would probably remove a few users who just level up slowly or only do reviews and no lessons and leave in a few users who quit during that time, but the vast majority of active users would probably level up within that time. If you’re interested I could see if I can generate a graph for that. :grin:

4 Likes

I decided to generate this graph since I thought it would be interesting to see the differences, but aside from taking out most of the higher-level free users and a lot of level 60 users, it doesn’t really change all that much about the actual distribution. The following graphs were made by only including users that either changed their level (either by leveling up or resetting) since November 2020, or joined after that time.

And a logarithmic one:

And one showing only users who are currently above level 3:

17 Likes

Thanks @BIsTheAnswer for your work. This definitely supports my earlier hypothesis

So if any of you want to get to level 60: regularly being part of the POLL thread is definitely going to help you achieve your goal

7 Likes

Could you add the horizontal grid lines, please?

3 Likes

That’s the thing. At some point, even with lifetime, you’ll stop using WK. Some insist on burning all, but I don’t hear that as a common goal in discussion. So, I wouldn’t assume that lifetime users=active users.

2 Likes

Yeah it doesn’t change the shape or analysis much, but the scale changed significantly.

Is there an easy way to re-stack the bars to be “that level or above”? Same info, but then you could simply put a ruler across at the top of your level’s bar and see how many people have been “in your shoes” at one point or another. (again, I’m expecting similar shape, different number scale, be interesting if I’m wrong)

1 Like

Done! I’ve updated both posts to include horizontal grid lines for the bar graphs and both horizontal and vertical for the line graphs.

1 Like

That would be similar to the total line in line graph I generated for the first post, but just for the active user dataset. I can just rerun the script on the other dataset and it should give me the graphs.

The results:


reach_active_log

13 Likes

Saw that other thread and realized I’m in a unique position where my sub is over, but I still have the account flair for paid <60.
image

I’ll try to refresh my login and get the other group.
image

2 Likes

Nice! Btw, I don’t know for sure, but I think people on level 60 are included in those numbers.

1 Like

Ah, so that’s why had the links like that. Whooops.

1 Like

I feel like WK is an anime, peope get killed off all the time. Even main character, the Game Of Thrones of Japanese learning websites

2 Likes

OP’s level probably says it all.

6 Likes

This topic was automatically closed 365 days after the last reply. New replies are no longer allowed.