Official Datamining Thread! (06/05/07)

[Stouter]

Try it on an element they are resistant to, or take off your elemental staff to test it. You might be past their INT already, which is why you didn't see an increase from INT. In any event, you're twisting my words by doing a test that appears to contradict me, even though it has nothing to do with what I said.

I'd rather have Mahatma pants when resists are an issue as BLM. For raw damage the NQ (new abj) would be better

Try it against a mob that is resistant to you as a blm in your HNM build. There's a point where adding skill will do virtually nothing for your resist rate, but loading up INT will make a huge difference.

 

[Uzor]

You did somewhat prove that 2 INT doesn't give 1 Macc, but I don't think anyone was debating that. You just proved that INT doesn't do shit when your skill is horribly under the mob's.

You were still below the mob's A+ level on skill, and I'm not sure you got up over to the mob's level with that Elvaan Rdm INT.

The mob's A+ skill in what? Elemntal Magic? Elemetnal Magic vs. Elemental Magic do nothing. If you mean A+ in INT, wait 'till I get home and I'll see if I can calculate their INT.

To be honest, I don't think any mob have A+ INT. I know Colibris have A in INT, but they are BLU, so that puts their racial INT to A, job INT to C and SJ INT to C. If I remember right, Greater Colibri have 96 INT, so I'd make a guess and say Flans have A racial INT, and due to being BLM/BLM, A/A job int, which would put them somewhere around 130 INT? Not sure, I'd have to calculate it when I get home.

What exactly is it you mean with "You were still below thee mob's A+ level on skill"?

 

[Bajaratt]

Because you didn't understand what i said. I didn't list the % for each type of resist, but the % of nukes that were 1/8 resisted or more, 1/4 resisted or more etc.. I did that to be coherent with the tests of sleep. When you cast sleep, you only know if you are full resisted or if spell lands. You don't measure the intermediate resist rates.

Let me rewrite the % for each type of resist :

..........Full dmg.....1/2 resist........1/4 resist......1/8 resist
Naked..38%............25%..................11%........26%
Macc...52%............16%..................18%........14%
INT......37%............15%.................21%........27%

I understand, you don't understand how percentages work. If you display overlapping results, percentages become meaningless; they only used for how a result compares in relation to the whole.

Your test is still not terribly trustworthy, since you're ignoring such a large amount of your results which could easily effect the overall conclusion. The very nature of the test of using a spell that you cannot accurately gauge it's effects undermines the results with that large of a margin for error.

 

[Ravic]

The only thing I want to add is why didn't Squeenix just make a Perdu Scythe with the Perdu Voulge's stats.

 

[pchan]

I understand, you don't understand how percentages work.

I know what % is you noob.

If you display overlapping results, percentages become meaningless;

I dont understand this sentence. Try to be clear. If you mean taking into account all the resist that were above 1/4, i'm sorry to say it is meaningful. When you sleep a mob, what you see is either total resist or sleep landing. You dont take into account 1/2 resist 1/4 resist etc, all those resists are considered success.

Your test is still not terribly trustworthy, since you're ignoring such a large amount of your results which could easily effect the overall conclusion.

What part ? If you don't trust my test do your own.

The very nature of the test of using a spell that you cannot accurately gauge it's effects undermines the results with that large of a margin for error.

I don't understand this part either. Do you mean it is stupid to test a mob that you know has high resistance ?

 

[uku]

The only thing I want to add is why didn't Squeenix just make a Perdu Scythe with the Perdu Voulge's stats.

QFT

 

[Lordseph]

They want more people to use GS instead of Scythe. Only explanation I can think of.

 

[Bajaratt]

I know what % is you noob.

Based on your posts so far in this thread, you don't.

I dont understand this sentence. Try to be clear. If you mean taking into account all the resist that were above 1/4, i'm sorry to say it is meaningful. When you sleep a mob, what you see is either total resist or sleep landing. You dont take into account 1/2 resist 1/4 resist etc, all those resists are considered success.

Percentages are only useful when they are independent of each other when you're listing results like that. The second table you made is infinitely more accurate than the first, because there is no overlap in the values being measured. If you can't measure the intermediate results why are you reporting them, and where are these numbers coming from? I'm betting somewhere in your ass.

Even if you could accurately guess how often the resists there's overlap in the data. If your second table is to be trusted, measuring the spells(while naked) with more than 1/8 resists would include the spells with 1/8, 1/4 and 1/2 resists, so you'd report 62% chance of being resisted; but then when you list the 1/4 or greater results, you'd give a 37% chance. The problem is, that you've inflated the chance of both resists because you're including additional and irrelevant data in your figures. If all you wanted to report was the chance of being resisted period, then 62% is the correct answer; but if you're rporting the number of 1/8 resists, 26% is the correct answer; anything else is horribly misleading and just flat-out wrong.

Percentages are only a worthwhile figure to cite when you're looking to the different compontents of something, so they have to add up to 100%.

What part ? If you don't trust my test do your own.

I don't understand this part either. Do you mean it is stupid to test a mob that you know has high resistance ?

My point was clear, at

Apparently you don't understand what you have to do for an actual, valid test or how to measure the accuracy of data. If you don't measure partial resist rates, or just don't include them in your findings, then you're throwing out a wealth of data that can drastically change your results. That's why I said the test is inaccurate, because your margin for error is so ridiculously high that your test results can't be trusted as proof of anything.

 

[Gregedor]

Not to derail the thread or anything guys, but this was the "Official Datamining Thread!" not the biggest dick competition for mages.

 

[CmdrFly]

Eh datamining's over, let the e-peen fire rage on. I be lurnin sutffz from this little debate. Just don't get too flamebait-y. Ain't nothin good gonna come of it should she wander over to the FB... we've already had our flame-that-ended-well for 2007.

 

[pchan]

Based on your posts so far in this thread, you don't.


Percentages are only useful when they are independent of each other when you're listing results like that. The second table you made is infinitely more accurate than the first, because there is no overlap in the values being measured. If you can't measure the intermediate results why are you reporting them, and where are these numbers coming from? I'm betting somewhere in your ass.

Even if you could accurately guess how often the resists there's overlap in the data. If your second table is to be trusted, measuring the spells(while naked) with more than 1/8 resists would include the spells with 1/8, 1/4 and 1/2 resists, so you'd report 62% chance of being resisted; but then when you list the 1/4 or greater results, you'd give a 37% chance. The problem is, that you've inflated the chance of both resists because you're including additional and irrelevant data in your figures. If all you wanted to report was the chance of being resisted period, then 62% is the correct answer; but if you're rporting the number of 1/8 resists, 26% is the correct answer; anything else is horribly misleading and just flat-out wrong.

Percentages are only a worthwhile figure to cite when you're looking to the different compontents of something, so they have to add up to 100%.


My point was clear, at

Apparently you don't understand what you have to do for an actual, valid test or how to measure the accuracy of data. If you don't measure partial resist rates, or just don't include them in your findings, then you're throwing out a wealth of data that can drastically change your results. That's why I said the test is inaccurate, because your margin for error is so ridiculously high that your test results can't be trusted as proof of anything.

I think that you are a total retard. If you actually check both table ( the second one being EASILY gotten from first one, at least if you know how to count) you'll see that i didn't forget anything. let me recapitulate :

First table

..........Full dmg.....>1/8 resist........> 1/4 resist
Naked..38%............74%..................63%
Macc...52%............86%..................68%
INT......37%............73%.................52%

Like i explained > 1/4 resist means counting all resist > 1/4 (meaning not the 1/4 one) cos if you do little math outside FF you now that 1/2 , 1 > 1/8 etc.
From this table you can easily obtain the 1/8 resist rate (Naked would be 100 - 74=26 etc). So no i don't hide shit. Doing such simple math as a 6 yearl old child could do you get the individual resist percent :

.........Full dmg.....1/2 resist........1/4 resist......1/8 resist
Naked..38%............25%..................11%.... ....26%
Macc...52%............16%..................18%.... ....14%
INT......37%............15%.................21%... .....27%

So no the second table is totally equal to the first one unless you need a brain.

If your second table is to be trusted, measuring the spells(while naked) with more than 1/8 resists would include the spells with 1/8, 1/4 and 1/2 resists, so you'd report 62% chance of being resisted; but then when you list the 1/4 or greater results, you'd give a 37% chance. The problem is, that you've inflated the chance of both resists because you're including additional and irrelevant data in your figures

1/8=1/8, 1/4>1/8, 1/2 > 1/8 1/1 > 1/8, Thus all the spells that were less (strictly) resisted that 1/8 totalalize 38+25+11=74%. Of course you need to count it the ones that were not resisted, because the first stable measures how well you land a spell. Here lands has the meaning of being less resisted than a certain %. You measure this for different %. So no there is no additional irrelevant data nowhere. You just need to learn math and what the sign > means. Oh and my percent add up to 100

Apparently you don't understand what you have to do for an actual, valid test or how to measure the accuracy of data. If you don't measure partial resist rates, or just don't include them in your findings, then you're throwing out a wealth of data that can drastically change your results. That's why I said the test is inaccurate, because your margin for error is so ridiculously high that your test results can't be trusted as proof of anything.

I did measure actual accuracy and i proved you with second table. My test are not random number i invented reading forum and such. I casted all those spell. Here is the sample.

1= non resit 2= 1/2 resist 4=1/4 resist 8=1/8 resist

mob= wamoura prince @ mount Z island

elemental skill naked = 242

GEAR 1 (Naked w/o staff)

2821221812|8184818181|1148118141|8182481112|1411821128|8114128812|2148821444|1282288881|1214212821|2212128121|

#1=38/100 #2=25/100 #4=11/100 #8=26/100

GEAR 2 (Magic accuracy + 15 only)

4281241218|1284144122|2144111188|4821421211|4411111111|1841141142|1142884421|1111481181|1221118818|1111211111|

#1= 52/100 #2=16/100 #4=18/100 #8=14/100


GEAR 3 (INT +30)

1421888818|2442841842|8128111281|8184121181|2141218811|8112482881|8181114142|8488288448|4112111412|4444114114|

#1= 37/100 #2=15/100 #4=21/100 #8=27/100

Feel free to verify it ^^


EDIT : i have now a sample of 200 of each gear which confirm the tendecy of the first 100 spells

 

[Stouter]

The mob's A+ skill in what? Elemntal Magic? Elemetnal Magic vs. Elemental Magic do nothing. If you mean A+ in INT, wait 'till I get home and I'll see if I can calculate their INT.

Wamoura Prince are level 79-81. At about 5 skill per level, that puts you at around 300 skill. It's not about elemental skill vs elemental skill, it's about your elemental skill vs their level.

 

[Bajaratt]

I think that you are a total retard.

And yet you're the one who barely seems to comprehend what a percentage sign is supposed to represent.

If you actually check both table ( the second one being EASILY gotten from first one, at least if you know how to count) you'll see that i didn't forget anything. let me recapitulate :

First table

..........Full dmg.....>1/8 resist........> 1/4 resist
Naked..38%............74%..................63%
Macc...52%............86%..................68%
INT......37%............73%.................52%

Like i explained > 1/4 resist means counting all resist > 1/4 (meaning not the 1/4 one) cos if you do little math outside FF you now that 1/2 , 1 > 1/8 etc.
From this table you can easily obtain the 1/8 resist rate (Naked would be 100 - 74=26 etc). So no i don't hide shit. Doing such simple math as a 6 yearl old child could do you get the individual resist percent :

.........Full dmg.....1/2 resist........1/4 resist......1/8 resist
Naked..38%............25%..................11%.... ....26%
Macc...52%............16%..................18%.... ....14%
INT......37%............15%.................21%... .....27%

So no the second table is totally equal to the first one unless you need a brain.

You should have tried explaining your process behind those numbers, because they made no sense at all initially. Now that you've explained what the hell you were doing, your method is stupid and the data you're presenting is almost worthless. Your first table is shit and completely wrong with how a percentage chart is supposed to work. Your second table is fine, but the first one is just stupid. Also, your value for 1/8 is completely misleading(probably poor English skills, but your stupidity could be a factor too), since you keep saying 1/8 resisted but apparently you mean a 7/8 resist and 1/8 damage remaining.

So when you're trying to find out the resist rates, but you add in the percent of spells that were not resisted at all into the percentage of your spells. That's ridiculous logic, since the whole point is the chance to get a resist, but you're adding in the chance to not get resisted at all; making the data worthless.

This is why your first chart is worthless and is completely wrong with what a percentage chart is supposed to display. The second is fine, because it displays all of the data that someone would want to know, AND THEN you can make the point of how you have a 74% chance of having more than 1/8 of the spells' damage hit; but combining the data together leads to the overlap that I was talking about.

In your first chart, you're combining the full hit, 1/4 and 1/2 resists in the second column, but in the third column, you combine the full hit and 1/2 resists. That means that the chance of a full hit and 1/2 resist are in both data points; but that additional data is completely irrelevant when you want to know the chance to get a 1/4 resist.

1/8=1/8, 1/4>1/8, 1/2 > 1/8 1/1 > 1/8, Thus all the spells that were less (strictly) resisted that 1/8 totalalize 38+25+11=74%. Of course you need to count it the ones that were not resisted, because the first stable measures how well you land a spell. Here lands has the meaning of being less resisted than a certain %. You measure this for different %. So no there is no additional irrelevant data nowhere. You just need to learn math and what the sign > means. Oh and my percent add up to 100

I'm guessing you failed math and science classes, otherwise you'd know how to make a table with percentages and what useful information would be.
[/quote]
38+74+63 = 175, looks like my math skills are just fine. I guess they just covered percentages in school for you this week, which is why you're so excited to use them and it also explains why you suck with them.

I didn't count on you using insane reasoning for your percentage values and trying to display completely pointless data that is easily gathered from an actually accurate table. The massive amount of overlap in your first table renders the data completely meaningless. Your first table is completely irrelevant, your second table is meaningful; at least it only took you two tries to do it right.

Your second table is fine, just stick with that and stop trying to prove how your idiotic first table had any value whatsoever.

 

[Stouter]

So no the second table is totally equal to the first one unless you need a brain.

Same information, but not nearly as useful. Reminds me of some of those drafting classes where they wouldn't let you fill in the useful dimensions because you already had the info to figure them out through 4 other dimensions already on the blueprint. I'd rather have those extra dimensions on there myself, whether I'm making them or reading them.

 

[mdkuser]

Wamoura Prince are level 79-81. At about 5 skill per level, that puts you at around 300 skill. It's not about elemental skill vs elemental skill, it's about your elemental skill vs their level.

actually they only have 2 different levels since thunder4 for example has 2 different dammage when unresisted.

pchan you should put the total damage dealt as well, since you wanna prove that magic acc rdm nukes better than int rdm.

 

[Stouter]

actually they only have 2 different levels since thunder4 for example has 2 different dammage when unresisted.

Puddings are only one level then, despite giving you different base exp? Not every level gives an extra INT.

 

[Lavi]

Its not all that uncommon for data to give overlapping information depending on the application. A common example is the lottery where they give inclusive odds and percentages for winnings. Ex: 50% are winners, 20% will win $10 or more and 1% will win big. Its expected that the user can intepret the data properly and it's not irrelevant or incorrect in any way.

However, the person providing the information must make the information clear to the reader so they may understand it. In this case it was not. BUT, considering the poster is a EU and non-native English speaker, the reader could use some common sense and common courtesy to ask for a little more information or to explain it better.

Same information, but not nearly as useful. Reminds me of some of those drafting classes where they wouldn't let you fill in the useful dimensions because you already had the info to figure them out through 4 other dimensions already on the blueprint. I'd rather have those extra dimensions on there myself, whether I'm making them or reading them.

There are various reasons for that. In my experience, drafting classes give the most basic of models which are made up of primarily simple geometric figures. Real world applications are rarely as simple and including all known data would be more than some space requirements can handle especially on a single drawing. I've seen a single drawing view have over a thousand dimensions without any overlapping or inclusive data. Needless to say, it was a rather complex drawing and sadly, it was just a solid base for a motor system and a mill.

There are also problems with over dimensioning objects. You can only have x equations and x unknowns or you are over specifying. Often over specifying dimensions can lead to computer problems and a computer cannot solve it and it will just toss or give an error with the last entered dimension. A simple example:
A=1
B=2
C=3
A+B+C=7

Which one is the error? If you excluded all extranious information that's not a given, then often the answer is very clear but to a computer, it has no idea which to chose and it's up to the user. Conversly, you are correct, it makes it harder for the viewer to understand the drawing, but a viewer has the capacity to make educated decisions whereas a computer cannot.

 

[Stouter]

There are also problems with over dimensioning objects. You can only have x equations and x unknowns or you are over specifying. Often over specifying dimensions can lead to computer problems and a computer cannot solve it and it will just toss or give an error with the last entered dimension. A simple example:
A=1
B=2
C=3
A+B+C=7

Which one is the error? If you excluded all extranious information that's not a given, then often the answer is very clear but to a computer, it has no idea which to chose and it's up to the user. Conversly, you are correct, it makes it harder for the viewer to understand the drawing, but a viewer has the capacity to make educated decisions whereas a computer cannot.

In that case the drawing is wrong. I've seen it too many times where you're looking for a simple value, and you have to find it by saying "ok it's this length plus this length minus this length times the cosine of this angle." If your extranious values contradict each other, then you've done something wrong, plain and simple. It shouldn't be up to the end user to determine what's right and what's wrong, because it should all be right. User friendly > minimal required information for any end user.

 

[Lavi]

In that case the drawing is wrong. I've seen it too many times where you're looking for a simple value, and you have to find it by saying "ok it's this length plus this length minus this length times the cosine of this angle." If your extranious values contradict each other, then you've done something wrong, plain and simple. It shouldn't be up to the end user to determine what's right and what's wrong, because it should all be right. User friendly > minimal required information for any end user.

All values are determined by the user as everything is user defined at one time or another. The issue isn't whether it's simple or not, but often many drafting softwares will not allow you to over specify. If you have two pieces of length with values for the total length and one of the partial lengths, many (all of the ones I use) will not allow me to modify or dimension the other value for that reason. They do not want the user to input the example I gave where the values contradict. Also, I would consider the fact that you did mention this being a class and you need to always keep your mathematics or more accurately, geometry knowledge fresh.

 

[pchan]

Stuffs
38+74+63 = 175, looks like my math skills are just fine.

Why the fuck would you add those number ? You look like you didnt understand anything.


Now I provided you with my test result. Feel free to give a personnal interpretation if you don't like mine, but at least don't qualify the test as wrong, because what i did is only note on a piece of paper 1/1 1/2 1/4 1/8 resist etc. And as for the terminology, 1/8 resist is when you do 1/8 of initial dmg, but i think you are the only one that didn't understand that. Resist is either 1/2 1/4 or 1/8. I've never encountered something else.