| Forum Index > BW General |
|
iFU.pauline
France1710 Posts
On December 26 2023 02:06 outscar wrote: The numbers are depressing tbh. I remember when Korea had like 20 or even 30k active. Why game with infinite potential isn't popular I don't get it... Because this VIDEO GAME with infinite potential is 25 years old?  | ||
|
XenOsky
Chile2356 Posts
| ||
|
bITt.mAN
Switzerland3693 Posts
You want to give a true picture. Giving a false picture would be bad - if the estimate is too big, it exaggerates the popularity and gives false hope. If the estimate is too small, it makes it look like the game is dying when it actually isn't. Your experiment is to sample player counts over time. Sounds simple to design a good strategy, but it's not. What: Like 2Pacalypse- advises, record player counts automatically every 5 min. Make sure you sample across all regions / launchers etc. to not miss anything. Ideally we'd also have more detailed stats like game duration or % that are ladder, or what % for each game mode, or what % for each map. But this isn't focused on the core question ('Popularity') and you can estimate those facts from replay databases. How: Set up a dedicated Windows computer, like an old laptop, leave it running 24/7. Install AutoHotkey and make a simple script that runs the launcher and clicks the the right buttons in the menus (or types the right letter hotkeys) to navigate to what you want. 'Read' the numbers. This is possible purely within AutoHotkey + Show Spoiler [How] + You can use AutoHotkey it to read out the pixel values of specific pixels based on their x & y coordinates on your specific screen. You can treat the numbers shown on the display similar to a 7-bar analog number display (might need more detail than only 7 points). Take enough example screenshots until you have one example of all 10 digits their font uses. Code logic filters depending on the combination of the pixels to conclude which number is shown. Most simple design would be to design it like a branching tree, where the first pixel you look at only appears in one of the numbers, then you can either stop there or eliminate that one and continue. You can use ChatGPT to help write all of this AutoHotkey code for you.) If that's too tedious, it could instead be possible to save screenshots with AutoHotkey (with the timestamp as the filename) to a synced folder, and run an automatic script to 'read' those numbers. Write those numbers to a synced file. You could append them as timestamped entries to a .csv file synced using Google Drive Desktop. Make multiple back-ups of this raw data. Why: From your responses and logic, it is clear that you have a nice idea, but you need more guidance to understand how to design a better sampling method. I suggest you do more research on this, such as by asking ChatGPT (preferably the paid GPT-4 - it is much much 'smarter'). In Data Science terms you are trying to sample 'Time Series' data to estimate different statistical measures. Currently it seems your bottleneck is human effort. But if you set it up once, then have this running 24/7 and automatically collecting data for you, you can instead get much more reliable results. As I started with, without reliable results, the value of the numbers would be 0 if not potentially damaging, so only do it if your sampling design is able to give reliable results. You want to estimate a realistic / true picture. You can't do that if you pick literally the worst time for people playing in Korea 🤦🤪. Don't pick 5am Korea time. But also don't arbitrarily pick 10pm Korea time. Don't make assumptions and guesses, because you don't know if those assumptions are true unless you collect more context. Just record everything in 5min intervals, only then will you have context to know for sure. You can only measure 'Total Number of Players', at different timesteps. By having lots more timesteps, you can get much more reliable estimates of things like: Mean, Median, Minimum, and Maximum Players. For the analysis, bin them in time intervals (5min, 10min, 20min, 60min, 120min, 6hr, 24hr, 7days, 28days, monthly, quarterly, yearly), perform time-series analysis (check if it is auto-regressive and/or has moving-average, check GARCH), and fit some models (simplest: linear fit + sine wave with periodicity of 24hr to model daily cycle + 365day yearly cycle). The best way to learn more about this is to copy-paste this entire thread into ChatGPT and it can explain anything, and can even help write the AutoHotkey code for you. GPT4 can even perform very good analysis of this sort of dataset for you automatically. I'll even do you a favor – if you set up a reliable data collecting setup like I described, PM me and I'll do the data analysis for you | ||
|
Sonic_md
Moldova278 Posts
First of all, thank you all for your comments, advice and suggestions. They helped me look at this table differently. Plus, for more than 3 weeks I tried to collect data as often as possible. Various thoughts, ideas, etc. appeared in my head. What to do next with the data from this table. I'll gather my thoughts together over the weekend and try to write. | ||
|
Sonic_md
Moldova278 Posts
When conceiving this collection, I did not take into account one important thing, which I understood, among other things, thanks to your comments. My data only shows the number of players on the server during the most popular time (from 22-00 to 23-00 local time). They only show the number of players at a certain time. And they cannot be judged by the total number of more or less active players on all servers. Regarding what they wrote here: collect with a bot or script. There are also certain difficulties here. For example, a person is on the server 24/7 (playing or just being there). The bot that will count will count it as different people. ALTHOUGH there may be one. So maybe my TOTAL number of fans idea isn't feasible. I will continue collecting data at the most popular time on the servers (22-00 local time). I hope this number will grow up every time | ||
|
Sonic_md
Moldova278 Posts
SC:BW Active Users/Games Played Data Collection (Table) | ||
|
Sonic_md
Moldova278 Posts
I have been collecting data for this table for 50 days. And here are some intermediate results: Average number of players at the most convenient time on all servers: 30,685 players. Average by server: West: 1508 East: 115 Europe: 619 Korea: 27062 Asia: 1391 Maximum total for all servers: 35595 - it was February 11, 2024 (Sunday) Maximum by server: West: 1905 (February 4, 2024 - Sunday) East: 165 (28 January 2024 - Sunday) Europe: 750 (December 30, 2023 - Saturday) Korea: 31575 (February 11, 2024 - Sunday) Asia: 1870 (6 January 2024 - Saturday) Minimum total for all servers: 24290 - January 25, 2024 - Thursday West: 1295 (31 January 2024 - Wednesday) East: 80 (February 8, 2024 - Thursday) Europe: 495 (February 14 - Wednesday) Korea: 21200 (January 25, 2024 - Thursday) Asia: 1035 (25 January 2024 - Thursday) I plan the next summary on the 100th day of collection. I hope the numbers will at least not get worse. It is clear that the most visited days are Saturday and Sunday. | ||
|
tankgirl
Canada453 Posts
On February 14 2024 23:38 Sonic_md wrote: Maximum by server: Asia: 1870(6 January 2024 - Saturday) Minimum total for all servers: Asia: 1870 (25 January 2024 - Thursday) ? | ||
|
prosatan
Romania8684 Posts
yes , i think this is a minor mistake , but still , nice data colection ! | ||
|
Sonic_md
Moldova278 Posts
thx, sorry. correct. 1035 - Asia minimum | ||
|
Sonic_md
Moldova278 Posts
The reason - may be Valentine Day or everybody watching BSL )) | ||
|
FriedrichNietzsche
92 Posts
And who knows how much more popular it could/would be if sponsors wouldnt push newer games so hard but go for quality (SCBW).. | ||
|
M3t4PhYzX
Poland4253 Posts
It's a shame, really. But these numbers aren't that bad anyway, considering everything. Especially for S. Korea. | ||
Xeofreestyler
Belgium6775 Posts
On February 15 2024 06:36 FriedrichNietzsche wrote: Regarldess of precise numbers.. there arent many games to begin with (chess comes to mind, CS maybe?) that are > 25 years old and still played at all let alone popular.. And who knows how much more popular it could/would be if sponsors wouldnt push newer games so hard but go for quality (SCBW).. Smash 64 is 25 this year. Smash Melee almost | ||
|
jzgjghjght
1 Post
| ||
|
Sonic_md
Moldova278 Posts
I continued to collect data for this table. And here are some intermediate results for the first 100 days: Average number of players at the most convenient time on all servers: 31970 For the first 50 days - it was 30,685 players. That is an average of 1285 more players every day. For the first 50 days - the average was 30685. For 51-100 days - it was 33280. That is 2595 more players on average every day.... In the first 50 days. The maximum on all servers combined was: 35595 - it was February 11, 2024 (Sunday) For the 51st through 100th days. The max was 40045 (April 6 - Saturday) In general, judging by the graph - the dynamics is still positive. There is no decline. | ||
|
Sonic_md
Moldova278 Posts
why can't I embed this chart? As a picture to the post. | ||
|
kAra
Germany1416 Posts
![[image loading]](https://i.imgur.com/26vgVbD.jpeg) | ||
|
[AS]Rattus
428 Posts
| ||
|
hamsterkoster
3 Posts
The stated goal of the site (Translated): This site is a StarCraft: Brood War game, event, player, map database, sharing and analysis site. The goal of the site is to collect as much data as possible, preserve the legacy of StarCraft, and provide global statistics to the StarCraft community. ![[image loading]](https://i.imgur.com/tkJSW1Q.png) Just look at how many games are being hosted each day! Over 300,000 on this weekend. It just came like 2 months ago so it's lacking long-term stats and features and better analysis but this is way better than spot-checking counts at a time during the day since it's automated. It would be nice to see if is updated to be able to break out data by server (since the data basically just represents Korea as Korea server is massive compared to the others). It also has some nice features like seeing which map was most popular in the past 1, 7, or 30 days. You are also able to download the map. The site has so far accumulated over 34,000 unique maps. There is a contact page, so maybe if there are some ideas you think might be good might not be a bad idea to reach out. Would have to be in Korean though. | ||
| ||
StarCraft 2 StarCraft: Brood War Dota 2 League of Legends Counter-Strike Super Smash Bros Heroes of the Storm Other Games |
|
Sparkling Tuna Cup
WardiTV Map Contest Tou…
Ladder Legends
BSL
CranKy Ducklings
Replay Cast
Wardi Open
Afreeca Starleague
Soma vs hero
Monday Night Weeklies
Replay Cast
[ Show More ] |
|
|
EPT
