LSF Statistics

[Tempus]

So for a web programming project I decided to gather and present some data about LSF's user registration numbers, mostly for my own curiosity. I figured I'd share the info.

Click image for bigger version.

Obviously 2015 is still in progress so the data for it is going to (and already has!) changed. From data collected in 2015 so far it seems total registrations for 2015 will be roughly 4,000. This is reflected in the extrapolation above in yellow. Here's a little more detailed breakdown:

Click image for bigger version.

I'll work on presenting things nicely later on, preferably in an automated way! Also I'll be adding more stuff later in addition to the number of registrations—number of recently active users within a certain time frame (e.g., of the 25,000 members how many have visited the site in the last 6 months?), percentage of accounts with zero posts, number of bots that've slipped through registration (I noticed some when scrolling through the database), etc. If you have a request for something you wanna see, I might do it. Just keep in mind it has to be possible with public user data. I can't analyse PMs (since they're private, obviously) and I can't analyse user location because LSF doesn't collect that data.

(Also, I know it'd be more efficient to ask Tom to do this but a.) this is a learning project for me and b.) he's probably got better things to do!)

[trooper6]

Could you collect which percentage of users self-list as over 18 vs. those who don't?

[Tempus]

Yes, that's definitely possible. So far I've only got data from the memberlist (mostly because it's more efficient to go through 418 pages than 25,000 individual profile pages), but I'll eventually write something that grabs info from every profile page.

[Kailoto]

This is really cool! Like trooper6 I'd be interested in the age distribution, and I'd also like to throw in posts per day for your consideration. I'm interested in what the local maximum would be (I'm assuming the absolute maximum would be zero, due to all the unused accounts), since that's a little more telling than the average or median.

[Tempus]

Glad you like it Posts per day would be interesting. I didn't pass high school maths, so I'll have to look up how to work out the local maximum when I get to that part!

---

On an unrelated note: it was fun watching all the user names go by when the script ran. There were a lot of funny ones. My favourite was:

• # 2843
  Username: CuriousHornyGuy
  Reg. Date: Fri Jul 10, 2009 5:00 pm
  Reg. Unix: Fri Jul 10, 2009 10:00 pm
  Post Count: 0
  Website:

I also ended up reading a fair few of old threads too, weaving my way to them via usernames that caught my eye. One of those threads was this one from 2005. Totally unrelated but hey... um, I don't have a conclusion to this sentence help. Cool to see members who still post today way back then! It also makes me feel nostalgic, strangely—strange because I wasn't even a member on this forum until 2013.

[Kailoto]

Glad you like it Posts per day would be interesting. I didn't pass high school maths, so I'll have to look up how to work out the local maximum when I get to that part!

Oh, you don't need to calculate it; just take the data and graph it, with posts per day on the x-axis and number of people on the y-axis. It would be a scatter plot (individual points), not a line graph, and the local minimum would be the highest point in a certain area. For example:

Here's a random scatter plot I found off of the internet - I circled the local maximum. Even though there are higher blue points, the circled one is the highest point in its vicinity. (Also yes, I do have a lot of free time.)

[PyTom]

For those who didn't see it, here's the visualization I made of the first ten years of forum activity.

https://www.youtube.com/watch?v=hLWeK7C5jag

[trooper6]

I hadn't seen that before PyTom...that is beautiful!

[PyTom]

Thanks. It's actually a program meant to visualize git repositories - who is working on what file in a program. What I did was to convert each forum into a directory, and each thread into a file, and then let it loose.

[Kailoto]

Thanks. It's actually a program meant to visualize git repositories - who is working on what file in a program. What I did was to convert each forum into a directory, and each thread into a file, and then let it loose.

That is quite the ingenious use of the program! I never would have thought of that, and it makes quite the spectacular result.

[Tempus]

@Kailoto: Oh neat. Seems like there'd be multiple local maximums (or maxima, I guess) then? I wouldn't be surprised if Google graphs (which is what I was using to make the line graph) has the ability to mark local maxima within certain thresholds or something.

---

Also, it's cool to see that visualisation of forum activity again. Pretty hypnotic.

[Tempus]

Just a quick little update for this thread:



I've been a little busy lately, but I'm hoping to come back to this soon.

[svarn]

um... I was going to say ...
"Wow 57% of ppl don't post ever and 42% of ppl post just once ?! That's so sad!", and then I was like ... wait ... something's wrong here ... thats 57+42=99
Check up for math buddy haha
Gotta make it to 5+ group fast, I feel bad about these stats.

[Tempus]

@svarn: haha, well, there is still the fractional parts of the percentages which bring it up to 100.

Regarding the stats, I wouldn't feel bad. Though I am curious to know whether this is an average participation rate for an online forum. I suspect that some fraction of those non-posters still lurk.

---

Another little thing I noticed while preparing my next script: currently LSF has 25,340 members, but the most recent member's ID is 32075, indicating that about 7,000 profiles have been deleted over the course of the forum's history. (Each profile has an ID associated with it. The first member is 0 or 1, and each subsequent member is one higher—2, 3, 4 ... 23464, etc. This number is used to link to your profile, such as in this link: http://lemmasoft.renai.us/forums/member ... le&u=19989 If you change the number at the end you get a different profile. For obvious reasons, the number associated with a particular user never changes.)

[svarn]

Ooooh never mind me sorry I'm stupid.
I assumed:
57,76% = didn't post
42,24% = 1-5 posts
19,93% = 5-50 posts
etc. Which already "adds up to 119,93%", You can probably see my mistake at this point > <