Martin Sleziak's room

« first day (2638 days earlier) ← previous day next day → last day (956 days later) »

Buraian

8:57 AM

Good day, I'd like some help for making an SDE plot of the following:
I want to know, out of all the users who joined MSE in a year, what fraction of users gained the privilege of becoming an established user relative to total user joined (1k rep) plotted as a function of time (unit=year).

Martin Sleziak

9:09 AM

in Buraian's not so secret hideout, 10 mins ago, by Buraian

I am in a bit of time crunch and that's why I hoped that someone with more expertise would have done it, but maybe sometime in the future. Thank you

Basically the same can be said about me. (There is stuff which is going on IRL.)

Another issue is that it is quite difficult to get what reputation users had at some point. (The database contains reputation in the Users table - this is the current reputation. But it seems that you're interested in something like reputation about one year since the user joined.)

You can find several posts on Meta Stack Exchange related to this issue: meta.stackexchange.com/questions/tagged/…

For deleted users, there is no chance of getting this information. For existing user, from the Votes table you can get at least a reasonable estimate.

Another issue is that it is a bit unclear to me what is actually asked.

For the benefit of others who stumble upon this chat conversation, this is related to this post on meta: What fraction of users gained the established user privilege over the years?

@Buraian Is what you're trying to get somewhat similar to this: Number of users who had reputation 1k at the end of 2021, 2020, 2019, etc? (And then divided by number of all users at the time.)

Buraian

I don't want it divided by all time

I want it divided by the number of users who joined MSE that year itself

So there could be a description of how many users gained the threshhold in a year relative to the ones joined

@MartinSleziak I appreciate your help.

I'm going to assume the number of deleted users are very less compared to the others

Martin Sleziak

I am not sure whether that's the case.

As I said, I will be rather busy (and still a bit unclear about what is asked here).

Buraian

I do understand. It would still be great if you are able to find the SDE query be it much later as well

Martin Sleziak

One thing you can do yourself - and you don't need any knowledge of SQL for that - is that from reputation leagues you can get some estimate of the number users who had reputation 1k+ at some point: stackexchange.com/leagues/69/week/math

Buraian

My hypothesize is that post EOQS the number of people who gained privileges after one year relative the amount which joined, is less than say a year back

Martin Sleziak

9:23 AM

@Buraian It seems to me strange that we're comparing two rather different things. Namely 1k+ users among all users and then we divide only by users who joined in the given year.

Buraian

I see

I didn't mean among all users

Martin Sleziak

So what is it you're actually want to count.

Buraian

Let me rephrase the problem, I'm not sure what would be the ideal query anymore

Martin Sleziak

You want to take all users who joined in 2020 and then the number of users with 1k rep at the end of the year only among those users?

Buraian

My goal is to check if less people are gaining 1k privilege post EOQS or not

yes

I want to compare that to users in say 2019, 2021.. etc

I think I need to think more of what would be the optimal query, as it is right now, it may not be good enough

give me a while

Martin Sleziak

9:27 AM

Seems a bit arbitrary (in a sense that if somebody joined on December 31, there is virtually no chance that they will have 1k at the end of the year). Something like checking how many users have 1k rep since the account was created would make a bit more sense to me.

Buraian

Ahhh

yes

that's the correct idea I think

Martin Sleziak

In any case, either of them will not be quite straightforward - as I mentioned, computing a reputation of a user at some point in the past isn't easy. (And we can only get an approximation.)

Buraian

I understand

Martin Sleziak

Anyway, at least now I roughly understand what is being asked. I definitely cannot promise that I'll be able to do such query. But I might try during the weekend if I have a bit of time.

And, of course, it is possible that somebody else will chime in and create some query for this. (There is no doubt that many users around here have much more superior SQL skills.)

Buraian

I'll ask around. Thank you for your assistance

Buraian

9:46 AM

@MartinSleziak I thought about it, but I realized it all sort of cancels out, those in mid year had a higher chance , and also those in start.

So, the original query should sort of give an approximation to the question

Martin Sleziak

10:30 AM

BTW I see that the question was reopened. If it's not closed again, there is some chance that somebody will post an answer with some queries: math.meta.stackexchange.com/posts/34603/revisions math.meta.stackexchange.com/review/reopen/5167

Just for the record, I'll add that there is a brief discussion in SEDE room: chat.meta.stackexchange.com/transcript/1223/2022/2/23

I certainly agree with this:

in Data Explorer (SEDE) on Meta Stack Exchange Chat, 17 mins ago, by Glorfindel

@Buraian it's difficult, since historic reputation information is not kept and you'll need ugly queries like this one to generate it from events (votes, bounties, etc.)

in Data Explorer (SEDE) on Meta Stack Exchange Chat, 16 mins ago, by Glorfindel

Maybe if you just concentrate on votes and don't pay attention to the reputation cap you can get a good approximation.

7 hours later…

Feeds

5:12 PM

Q: What goes into site sponsorships on SE?

In 2018 we introduced site sponsorships as a revenue stream to generate income that would go back into Stack Exchange. Now that we’ve been doing this for a while, and in response to community questions, we thought that it might be helpful to share more about the process. Who can sponsor a site? T...