lol, i'm speechless at yukito, there were actually 10 of you who did such a thing for a custom game in warcraft, you guys were really enthusiastic weren't you

anyway, since you've done all the statistical analysis, i guess i should believe you that when a hero is hitting a creep having a quarter of its hp left, the damage dealt is skewed and no longer a normal distribution, but nevertheless it is still random, just not normal anymore


he just said he did with his friends long time back when they were fanatic about dota, i used to be pretty fanatic too about it, so i guess it's believable

and about "How do you note down the hp of the creep when the fighting between creep are so intensed?", i can answer for him, it wasn't a normal game where 5 people vs 5 people, him and his friends created a game solely for this experimental purpose, and when you're doing such an experiment, of course you don't whack the creep which is being whacked by other creep, but whack the creep which wasn't attacked, therefore the other people could note the damage dealt each time a hit is made, and they could always look at the replay to avoid mistakes in noting down..

50 is enough, each creep has around 800hp as he stated, so chaos knight would need an average of 15-20 hits to kill a creep, ie 50*20=1000 samples, that's sufficient

Do note that he's doing it in a controlled environment (no enemies etc.), so recording the health of creeps is pretty easy (Just target the one creep that no one else is attacking)

Though I agree that a sample of 50 is rather small, justifying it by saying that "100% is full, and one creep represent 2% of the data" doesn't make any sense. Sample size is determined based on the total number of average kills per game, which meant that restarting each game isn't necessarily scientific, but doesn't affect the sample size. In the real world, sample sizes are often between 0.01 (TV ratings) - 25 (Small experiments) % of the data. So, the reasoning behind the sample size is based on the average number of creep kills a player makes in the first 5 minutes, which would actually make the sample size too big. But, we assume its the overall picture, so the average creeps kills a player makes is around 100-200, so 50 is a pretty decent number.


Anyway, there are a dota fanatics out there who are indeed willing to carry out crazy experiments like this.

BTW, yukito, you didn't post the tables and graphs for your experminent

Lolz yukito, show us the prove!!! ROFL... I am quite pretty sure yukito wont do this... he still have his honours project to finish... or he took this as his honours project? Lolz. Topic: "Statistical analysis of DOTA: To discover the randomness of DOTA by last hitting creeps."
Btw, you said: "Do note that he's doing it in a controlled environment (no enemies etc.), so recording the health of creeps is pretty easy (Just target the one creep that no one else is attacking)."
So every creep is full HP then? How to you judge the random-ness of last hit kill with creeps of full hp? The lower HP ones are being attacked by amass of opponent's creep, which the figures jumps like crazy.
And btw, if you attacked a full hp creep to be killable with last hit, the opponent's creeps AI will immediately target that particular low health creep, making it unjudgable by whether it's killed by you or by them.
And by lvl1, it's pretty hard to do this...

I believe that by default, creeps do not discriminate against targets except when an enemy targets an allied hero. Only towers will target creeps with the lowest health, normal creeps do no such thing, otherwise those ppl who go for the necro/druid will often aggro the rest of the creeps the moment they attack, which doesn't happen. It is possible for a hero to land ALL the hits on a single creep and kill it in a single wave. Worst comes to worst, just engage the creep wave minus your own creeps (Or use the extra heroes as 'targets')

In a normal encounter between friendly creeps and enemy creeps, they often focus their attack on the first creep they encounter, regardless of how much HP they have. Besides, creeps and level 1 heroes attack ridiculously slooooow, so recording these values is not as far fetched as it sounds... Heck, they could be really anal and use FRAPS to record the game.

Now I see what you mean, but the way you said the CK's stun damage is fixed, you are saying as though there is a fixed damage instead of 1-200. It is better that you see each stun as an invididual probablity, instead of taking 100 or whatever stuns.
Funny how you brought in the normal distrubtion for this. And by theory, n>30 is enough.
I am not disputing what you have explained, but I am disputing the way you put your replies before that. 99% of people who read your CK's stun reply would most probably go " eh? does that means there is a fixed damage for the stun and not 1-200? ".
=)
Edit: I still doubt it though, I'll try and ask around.

This post has been edited by Sesshoumaru: Mar 30 2006, 04:36 PM

1000 samples is usually a number that is, IHMO, relatively significant. This is shown that a normal distribution can be achieved simply by the data from the table 1. Remember that it is the difference in both graphs that is the key point and demonstrates and non-randomness of damage dealing.


When checking random-ness, normal distribution is usually what comes into my mind.


So there are 5 heroes right? 2 buy flask of sapphire water, 3 buy four gaunlets of str (the 12 additional dmg is welcomed as there are more variables to etst with). Basically what we do is we go for the next waves of creep if possible, those where creeps are not involved and would not interfere with the results. Flask to heal.

walabies, remember the time when I was promoting the greatness of DotA and you were just lying on the bed promoting normal warcraft and saying how lame was it? It was at that time I did the test

P/S now doing watch movie. I have something else to say but I would tell u guys later.

This post has been edited by Yukito: Mar 31 2006, 03:58 PM

you mean like sniper's and lina's 'invisible black dot' from scattershot(v5) and dragon slave(both versions)? if that's the case.... then i really have nothing to say but this: i surrender to dota's uncoolness.... *sigh*

(btw i forgot what i have said about the random thingie lolx )

too bad i was looking at him and i found out that Kraken Shell did NOT at ALL, cancel my stun.... it's bloody impossible..., not the way i see it in v5 =.=



This post has been edited by sanjikun: Mar 31 2006, 12:59 AM

Dude as I said the results were separated into two tables based on the creeps HP. Refer back to my post.


I wish to remind you guys that the notion possessed by AI to "attack lowest health creep" is only implemented after v5.67, the version I tested on. Before that, the AI was really simple, frag all enemies they see, its a no-brainer for them.


Don't worry, I am not THAT crazy yet

I think you guy start talking shit.. is all this dicussion important or it is because you guy are bored with dota and start making nut thing. Quit dota then..hahaha "Dota is suck"

So, to counter our "shit" you provide even less useful information. I mean, aside from your b****ing, you haven't contributed (Positive or negative) anything at all.

It was when 5.84 vers, when DotA is really... imxxlxxxe compared to normal game. I just really ca accept with few heroes dominating the game. Started at ver 6.
Well now I am standing in the company doing work. No time DotA.

A ten year old kiddo challenging the world?

Peace everyone...

I just noticed that Ursa can't benefit from Battle Fury's cleave attack

"

Maybe it's because of the fury strike...

i think i read sumwhere dat ursa is a ranged hero at 100 range...dats y bfury doesnt work on him

He's ranged in normal attacks, melee when on ultimate.

After being through DotA since the old times, the balancing issue is always there, no such thing is "now is more balance than last time". Its how you see it.


No such thing as few heroes dominating the game, except for clan matches. And its still the case in ver 6

just 1 day then we got another version of dota

Ver 6 clan matches have different combos now, just for your reference. New heroes add and old heroes adjusted. Not like the dull old same o hero in ver 5.

Version 5 has quite a number of comboes available, if you check out the replays of tournaments To say they are dull, is inaccurate at best.

In v6 clan matches do you guys think hyperstone > mask of madness for Sven?

http://www.dotaportal.com/forums/index.php?showtopic=22527

Rofl 10 minutes game, now that's what i call teamwork.

actually i dun think any competitve matches are bored ... cos they are organized hehehe i like it ...