EntropiaTracker.com Official Thread

[Starfinder]

HAHA... solved the mob name issue.... here's the code.. in ASP.. now to convert it into C# so i can include it in the bot..

element = trim(rs(0) & " " & rs(1))
	element = ucase(element)
	testword = ucase(testString)
	
	'Multiply result
	if len(testword) > len(element) then
		longWord = testword
		shortWord = element
	else
		longword = element
		shortWord = testword
	end if
	matchNumber = 0
	
	
	'We need to run through the longest word. If we do it with the shortest word testing for "Allophyl old" will generate a high hit on "Allophyl old alpha" - we dont want that.
	for i = 1 to len(longword)
		'sLen = search length. First we just test for 1 letter.
		sLen = 1
		'We asume we find something
		found = true
		do while found = true
			'We can only continue the loop until the total value of sLen and start index is less than the string. (otherwise you get index out of bounce)
			if sLen + i >= len(longword) then
				found = false
			else
				'Look for the "substring" in longword in the short word. (with an incremented distance to look for)
				if instr(shortword,mid(longword,i,sLen)) > 0 then
					sLen = sLen + 1
					'Find position diffrence.
					posInLong = instr(longword,mid(longword,i,sLen))
					posInShort = instr(shortword,mid(longword,i,sLen))
					
					if PosInLong > posInShort then
						diff = PosInLong - posInShort
					else
						diff = posInShort - PosInLong
					end if
					
					'We only accept letters that are placed 1 or 2 chars away from the "right place".
					if diff < 3 then
					matchNumber = matchNumber + sLen
					else
					'if the substring we have found is further away. We really hate it.
					matchNumber = matchNumber - (sLen * 2)
					end if
				else
					'If the string isent in the shortword then we havent found something. i then becomes 1 again.. and we loop again.
					found = false
				end if
			end if
		loop
		i = i + 1
	next
	'response.write element & " score : "  & matchNumber & "<br>"
	
	'Now we have a total score of the current element. Check if its a better score than the previous score - if it is, this is the best match.
	if matchNumber > mobMatchNumber then
	mobMatchNumber = matchNumber
	mobName = element
	end if
	
	
	
	rs.movenext
	loop
	rs.close
	set rs = nothing
	
	response.write mobName & " won with a score of: "  & mobMatchNumber & "<br>"

[Starfinder]

with the tweaks... the program is now up and running again... and i bought extra coolers for my server so i think it wont overheat as much as it did before.. (30 hrs of non stop PE was hard on my baby)...

[falkao]

oh man, this is what I was looking for for ages. Hope we'll get some nice stack of data.

+rep

[falkao]

and here what I get from the data n = 50:

first of all a Kaplan-Meier estimate of the survival function:


Click to enlarge

it seems to follow once again an exponential distribution. So doing a regression I'll get an exp with mean 50. Since the mean of an exp is the mean and we have to add 50 due to global start, mean of the observations it is the mean of the observed loots as well.


Click to enlarge


there is some overestimation in the beginning, which I have to further analyze.


Click to enlarge

All in all, mean from Exp is 50 PED and globals start at 50, which means that a global will have in mean a value of 100 PED. Median will be 70 PED, hence 50% of globals will be between 50 and 70 PED.


So let's wait till hof's get recorded.

[Starfinder]

select avg(lootValue) from loots
= 145 PEDs .. (with a couple of big ones)

[falkao]

Originally Posted by Starfinder

select avg(lootValue) from loots
= 145 PEDs .. (with a couple of big ones)

max was 440 and I had 50 entries. Now I see there is one Hof more. So maybe you have more data as is displayed.

Edit: I see, I get the last 50 or soemthing entries. Only when I filter I'll get more. However, only aggregated, so not very usefully when analyzing distributions.

Furthermore, I think we should distinct in globals and hof.

Is there a possibility to add mob HP?

I did it manually and there is no correlation between HP and loot. This means that globals are not higher with higher HP.

Edit:
Max expected value for a global in the case of an exp dist would be 395 PED with a prob of 99%. So this can be used to distinguish between globals and hofs.

[Starfinder]

I think there is about 4K entries in the db now...

[genious]

hmm can't seem to find my team globals
teams in the list?

[falkao]

Originally Posted by Starfinder

I think there is about 4K entries in the db now...

indeed, so quite a lot. Calc avg for loot below 400 and you'll see you'll get a mean about 100. It might be that it is between 95 and 100.

[alh]

damn, you beat me to it

+rep for publishing it.