Accuracy Stats?

Discussion in 'PlanetSide 2 Gameplay Discussion' started by a-koo-chee-moya, Oct 8, 2014.

  1. Borsty

    Unfortunately, to determine which weapons are really more accurate than others, you would need to conduct a series of tests with hundreds of people. What you need is a test course, where the "test subject" encounters all the relevant situations, including small and large battles, good and bad FPS & pings, fighting outnumbered or outnumbering, day/night, situations involving vehicles.... you see, very very difficult and we don't have the tools to stage that.
    Of course, all the data is in theory available from the live servers. We don't have the tools to properly analyse them though, and I'm afraid neither has SOE. It would be a very nice job for a dedicated data miner this.
    Until we see those data, the whole discussion and pulling arbitrary numbers like overall infantry accuracy is totally pointless.
  2. Iridar51

    If you're interested in weapons themselves, here are the stats. And attachments. You're welcome to compare them on your own time.
  3. Zotamedu

    There are a bunch of weapons that are identical for all factions. Compare the accuracy of those and you'll get a fairly good idea on if there's a significant skill bias between the factions. Note that in this case, skill only includes player accuracy.

    For instance, the accuracy for the default dumbfire launchers are about as identical as one can expect.

    Shrike: 53.7% +- 8.4783% and 2.9% +- 1.4693% HS
    ML-7: 54% +- 8.5217% and 3.1 +- 1.751 HS
    S1: 54.8% +- 8.611% and 3% +- 1.5991% HS

    Within one percentage point with a standard deviation of 8.5 percentage points is about as even as you will get.

    It's the same story if we look at other weapons like the bursters, MAX punch, knives, grenades, bolt action sniper rifles and all the common vehicle weapons. I can't find a single common weapon with a decent amount of users that has a statistical significant difference between the factions.

    You haven't shown anything. You made an unbacked claim, that is not the same as showing something. I showed something and that's that there's no apparent difference in accuracy between factions when comparing identical weapons.
    • Up x 4
  4. a-koo-chee-moya

  5. CipherNine

    So how do you explain the fact that VS accuracy on NS weapons is much lower than average accuracy(which includes NS and VS-specific weapons) relative to other factions?
  6. Zotamedu

    I assumed you had a basic grasp of statistics. Oh well. The stats I looked at were all guns that are identical and they all had a couple of thousand users. The average is taken over the last month. Since all factions play the same game and the same continents, they will all encounter the same fights over the coarse of a month. The warpgates rotate frequently now as well, it's not like back at launch where the warpgate location might have skewed the stats. So we have the data and the data show that's there's no difference in accuracy between the factions when they are using the same weapons.

    Have a look at the central limit theorem. There is no reason to assume that we are dealing with anything other than a normal distribution here. If I could be arsed to scrape the data needed from the API, I would do some Student's T tests as well but it would take too much time relative to what it would give me. Mainly because I would have to learn how to scrape data via the API. If I had the data formated, then it would only take a couple of minutes to do the maths itself. Then we could compare overall faction accuracy as well.
  7. Vaphell

    goddamnit, stop answering people with other people's posts, formulate your own opinions.

    You don't have to tell anybody with half a clue about statistics that it's a tricky thing to answer questions about multidimensional reality.
    But you've made a strong claim, that the distribution of "skill" among thousands of players is not necessarily equal. Default situation is that "Skill" of people should average out if there is no selection bias, which as far as i am aware is true. Your claim was disproved by the statistic on mirror/NS weapons so we are back to square 1, ie 18% vs 21%. Feel free to poke holes in it, we will learn from it.

    Is the 18 vs 21% acc a problem? Maybe, maybe not, but that disparity warrants at least a second look. You apparently have your faction loyalty offended or whatever and refuse to even look at the numbers that represent fractions of reality and may indicate a problem.


    Is there any data on weapon classes, eg how LMGs or ARs as a whole perform? Maybe it would narrow the disparity down.
  8. a-koo-chee-moya

    Your NS weapon picture is also relatively incomplete due to the 20 ish NS weapons that have no faction comparisons
    You only have a few real comparisons due to the different nature of melee weapons and grenades. Just becuase factions are equally skilled with them does not mean that they are equally skilled with auto weapons.
  9. a-koo-chee-moya

    Hey, I thought that his idea was a good one, and didn't bother to reinvent the wheel.
  10. Zotamedu

    I would also like to add to people who are not that into statistics that the numbers 21 and 18 in themselves are rather meaningless since we don't now the accuracy of the numbers. (see what I did there). A reason that I like Das Anfall is that they tell you the standard deviation. If we have 18 % vs 21 % with a standard deviation of 10 %, then it would be pretty safe to assume that they are probably not statistically different. But if we have a standard deviation on 1 %, then there is most likely a difference. In any case, we would need a little bit more data to tell the difference.

    Anyway, I agree that the numbers someone posted are interesting if they are true. It's at least an indication that there might be something to the claim even if it's not a confirmation.
  11. a-koo-chee-moya

    The problem with Dasanfall sometimes is that people post stats without looking at the deviation. Thus, people post Q4 KPUs with deviations almost as large as the acutal number.
    You are right though, that the poster did not actually post the deviation which I would assume to be pretty large due to the number of weapon's accuracy needed to be calculated.
  12. Zotamedu

    Notice how I wrote significant number of users. The Helion you use as an example has only one user per faction. It's not an average, it's the accuracy from two users against each other. The Drake has 2 users for TR and 6 users for VS.

    I only compared weapons there have stats for each faction and that are the same for all faction. Notice how I wrote common vehicle weapons.
  13. a-koo-chee-moya

    1.I did compare common vehicle weapons....

    2. Lol I can't beleive I read those 2 wrong, sorry for the waste of time. I was reading the DPS... Too much scrolling I guess... Should've noticed the similarity. Walker is still out there though....

    3. Those NS weapons without faction comparisions?
  14. Zotamedu

    Noticed how I used deviation and how I completely ignored KPH and KPU.

    Your assumption is not very relevant because you don't seem to have the basic understanding of statistics to back such a claim. Your assumed that there was a skill difference between the factions and in game numbers could show no such difference. Why would your gut feeling be right now?

    If we just scroll through the guns, it seems like most infantry weapons fall around 5 % standard deviation. Shotguns seems to be more like 10-15 %. So a 5 % standard deviation on a 18, 18, 21 accuracy might not be significant. What makes it interesting is the fact that NC and TR are so close. It might be nothing and it might be something. It would be nice to see some proper data mining on the subject. Not that it really matters. Also, the answer to your question is already pretty much answered when we look at the overall stats on the guns. There was a couple of interesting posts on the first page about it.
  15. Gammit

    I've been wondering the same thing, OP, as I almost always lose to VS anywhere between 10m - 60m unless they don't see me. Granted, I probably suck, but I don't I'm that bad. Hell, LAs and medics routintely bring me down when I'm running Heavy.

    I'm probably going to log into my VS character soon and pencil down some stats of the weapons that keep popping up on my killed-by screen. The only one I can remember off-hand is the Orion.