Randomerest Question Thread III: This Time It's Random

[Drunkenmonk]

114 total http://nullvector.com/enhancing.php

[Deband]

Hiya, I just got a new PC and i remember there were some threads on locking down Internet Explorer and how to set up Noscript. Can anyone point me in the right direction?

Also I can't seem to find my TOAU disc/box. Should i just get that collection thingy?

[stewie]

Hiya, I just got a new PC and i remember there were some threads on locking down Internet Explorer and how to set up Noscript. Can anyone point me in the right direction?

Also I can't seem to find my TOAU disc/box. Should i just get that collection thingy?

GUIDE: Protecting your web browser. - Windower

[Necromage]

Dumb question, how do I change fonts in FFXI? I'm using the default and the letters are all screwed up ( even better question is how do I make the letters not get cut off randomly? )

Anyone?

[Edrakali]

Can anyone confirm that you can get -5% singing cast time on BRD fey dagger?

[Malinn]

Mmmkay, this is going to sound really stupid, but I've been working on ranged set for my Samurai, and I'd like some pointers. The thing is, not for sidewinder. I've got that covered, believe you me, I'm talking about good ole fashion Ranged attacks. Why? It's so very situational, but sometimes I've found myself wishing I could just take pot shots at mobs when I need to keep my distance. It's mostly a pet project for fun, I have no interest in leveling RNG (although I capped my marksmanship on war, haha) I'm not sure what would be ideal for regular ole DoT ranged attacks on a Samurai.

Hagun/Soboro (if I can't get in at all, Fire staff)
Shigeto Bow (+1 is outta my price range atm, Master Caster's bow is an option)
Optical Hat
Kirin's Osode
Diabolos Torque (don't ask)
No earrings yet (I lost my beater's/vision) what would be best for dot?
Behe ring x2
Amemet mantle+1
Seiryu's Kote
Saotome Koshi-ate (Go go SAM belt)
Galliard Trousers (this is a problem slot, the -agi bugs me, even as Mithra)
Enkidu's Boots (Another problem slot, agi+3, r.att +4 is nice tho)

Anyways, this isn't something really important, I just wanna optimize my fakeRNG for shits and giggles. Oh, and I'm doing the same for RDM/RNG for laffs. Bwhahahahah.

Input greatly appreciated.

[Thorny]

It doesn't matter how long your delay is with rune chopper. The haste will be active for 3 seconds, then inactive for 3 seconds, if you happen to swing while it's inactive you don't get the haste bonus. Even if your delay is 10-15 seconds, you'd still have the haste activating and deactivating every 3. Also, sanction refresh will do the same thing without making you tp in a godawful bodypiece.

[Draylo]

Why is there a random apkallu strolling through pudding camp... don't think I have ever seen this before lol is it a quest?

http://img.photobucket.com/albums/v4...801_100353.jpg

[Thorny]

Why is there a random apkallu strolling through pudding camp... don't think I have ever seen this before lol is it a quest?

If you trade a npc apkallu certain types of fish, they'll follow you for a while. As far as I know they don't do anything besides follow, but meh.. someone probably was taking it for a walk.

[Gere]

Otherwiki says that having them follow you while you kill apkallu makes you gain hate faster and that they eventually turn into mobs and attack you.

[Suiram]

I have a statistics question, I guess. If you are trying to get an idea of how common a particular event is by collecting data from a variety of sources (let's assume under equivalent conditions), does that data get skewed if each person stops collecting it once the desired outcome is reached? If so, how badly, and is there a way to correct for it?

For FFXI examples, like let's say I wanted to know the drop rate on some item that people want multiples of, like tree cuttings. So I have 10 people, maybe on different servers, but they all have the same level of treasure hunter and they kill the same goobbues, and report "I killed X gobbues and got Y tree cuttings" I could take (I should think) each of their data and combine them all together to get a better measure of the drop rate.

But what about something like a peacock amulet, where people kill until they get a drop and then when they get it they stop. So you get a bunch of reports like "I killed Argus and got it on the Xth kill." I'm assuming that will skew it somehow since you don't get info on how many would have dropped if they'd kept killing after. Is this right, or is my logic here flawed? And if not flawed, is there an easy way to correct for it?

[Deftscythe]

Is FFXIAH safe to use at the moment? I know it's had problems with viruses in the past.

[_chuchu]

Never had an issue while using noscript and adblock. Just don't allow scripts

[Thorny]

I have a statistics question, I guess. If you are trying to get an idea of how common a particular event is by collecting data from a variety of sources (let's assume under equivalent conditions), does that data get skewed if each person stops collecting it once the desired outcome is reached? If so, how badly, and is there a way to correct for it?

For FFXI examples, like let's say I wanted to know the drop rate on some item that people want multiples of, like tree cuttings. So I have 10 people, maybe on different servers, but they all have the same level of treasure hunter and they kill the same goobbues, and report "I killed X gobbues and got Y tree cuttings" I could take (I should think) each of their data and combine them all together to get a better measure of the drop rate.

But what about something like a peacock amulet, where people kill until they get a drop and then when they get it they stop. So you get a bunch of reports like "I killed Argus and got it on the Xth kill." I'm assuming that will skew it somehow since you don't get info on how many would have dropped if they'd kept killing after. Is this right, or is my logic here flawed? And if not flawed, is there an easy way to correct for it?

If it's true random generation, it shouldn't have a memory. As such, you can just look at the next player reporting data as a continuation of your data set. Killing 15 Arguses before you get your Peacock Amulet does not raise the chance of it dropping on subsequent Arguses(Argii? Fuck, who knows.). Your data is accurate provided the event in question is a true random chance and not something that gets changed in any way between occurances.

[Yugl]

If it's true random generation, it shouldn't have a memory. As such, you can just look at the next player reporting data as a continuation of your data set. Killing 15 Arguses before you get your Peacock Amulet does not raise the chance of it dropping on subsequent Arguses(Argii? Fuck, who knows.). Your data is accurate provided the event in question is a true random chance and not something that gets changed in any way between occurances.

Just pointing out the bold in case you neglect it since this *will* have an impact on your data (If the chances of Argus giving an item is dependent on previous events, it will have an effect).

[pchan]

I have a statistics question, I guess. If you are trying to get an idea of how common a particular event is by collecting data from a variety of sources (let's assume under equivalent conditions), does that data get skewed if each person stops collecting it once the desired outcome is reached? If so, how badly, and is there a way to correct for it?

For FFXI examples, like let's say I wanted to know the drop rate on some item that people want multiples of, like tree cuttings. So I have 10 people, maybe on different servers, but they all have the same level of treasure hunter and they kill the same goobbues, and report "I killed X gobbues and got Y tree cuttings" I could take (I should think) each of their data and combine them all together to get a better measure of the drop rate.

But what about something like a peacock amulet, where people kill until they get a drop and then when they get it they stop. So you get a bunch of reports like "I killed Argus and got it on the Xth kill." I'm assuming that will skew it somehow since you don't get info on how many would have dropped if they'd kept killing after. Is this right, or is my logic here flawed? And if not flawed, is there an easy way to correct for it?

Mathematically speaking, you are getting a random variable N which is the number of NMs until the fist drop. Since each kill is independent to the next, the expected value of N is the inverse of the droprate -exactly-. Now if you want to approximate this expected value you will need a lot of different players...

[CDF]

(1) Assuming a fixed drop rate, random sampling, and no "reporting bias" (major assumptions), if your sample size for any one sample does not depend on whether PCA drops, then there should be no real worries about pooling the data.

(2) But, if the estimated drop rates for PCA are reported only after PCA is obtained, the "natural" pooled estimator is asymptotically unbiased (and normally distributed) but is biased for finite n, where n is the number of individual samples used to construct the pooled estimator.

Pooled drop rates tend to be some combination of (1) and (2), so intuitively speaking any bias that manifests in (2) is blunted.

[Malinn]

What, no love for my pew pew pew bow Samurai question?

[Thorny]

bow sam is bad and you should feel bad

[Beave]

What would be a decent body to use on brd assuming it's a) just a job you'd hop on to sv madrigal or b) merit or other small event with? Don't feel like dropping 5m on manteel, and farming periqia for points is going to be rough with salv 4-5 times a week. =(