Statistical Map Research

-Player level: 77
-Player exp before running the map: 666,266,754
-Player exp after running the map: 669,098,995
-Player item quantity bonus %: 0
-Map name: Punishing Overgrown Ruin Map of Hordes
-Map level: 66
-Map +size bonus % : 0
-Map +monster pack size bonus % : 29
-Map +magic monsters bonus %: 0
-Map +rare monsters bonus %: 0
-Map is a maze: no
-Map has the "2 unique bosses" affix: no
-Map has a inhabited by special affix: 0
-Map +item quantity bonus % : 12
-How many maps and which level maps dropped: 0
---
-Player level: 77
-Player exp before running the map: 666,561,140
-Player exp after running the map: 668,713,994
-Player item quantity bonus %: 0
-Map name: Labyrinthine Dungeon Map of Commanders
-Map level: 66
-Map +size bonus % : 0
-Map +monster pack size bonus % : 0
-Map +magic monsters bonus %: 0
-Map +rare monsters bonus %: 28
-Map is a maze: yes
-Map has the "2 unique bosses" affix: no
-Map has a inhabited by special affix: 0
-Map +item quantity bonus % : 3
-How many maps and which level maps dropped: 1 Map lvl 66

I made some statistic about map drops - https://docs.google.com/spreadsheet/ccc?key=0Au90_YPp2dFJdDQ3dXVJakZGVEJBeXItTkJMaEt3dnc#gid=0

It could be useful for you

Thanks I'll see how I can integrate that in my data :)

I'm curious what does your modifier and modifier+qty data mean?

edit: Big thanks, I added all your data in my spreadsheet

"

Zoomtag wrote:

Thanks I'll see how I can integrate that in my data :)

I'm curious what does your modifier and modifier+qty data mean?

edit: Big thanks, I added all your data in my spreadsheet

Ok, that would account for the swath of rows at the end with no xp data. Ran a few maps last night in a 2-man party. Been leveling alts and busy with other stuff lately.

I was thinking about discontinuing this project.

Unfortunately we're not gathering enough data to draw reliable conclusions. The problem is that patches modify maps, map drops etc every now and then and we cannot accumulate enough data between two patches. To get enough data, it would require an automated client sniffer (Wowhead Client style) but I think PoE loot mechanics make it impossible to automate.

On top of that, there are theoretical statistical issues that I'm not sure I can figure out by myself; ie the multicovariance. Google spreadsheet also seems to have its limitations, missing some important statistical formulas, getting slow/memory issues as the sheet grows above 500 entries...

Finally I dont think we're producing any remarkable results other than confirming intuitive ones i.e. maze = wow, item quantity = meh ...

What do you guys think?

The recent changes to map drops don't likely effect the data collected much at all. Most of the rows are for 66-68 maps. Very few L70+ data points and these are likely the ones that were buffed. Future patches are likely to tweak stuff, but that doesn't necessarily invalidate the existing data in a material way.

As for google docs... well the analysis could be done using a program better suited for that purpose. The google doc could then be used only for recently collected data and posting the results of the analysis, updated periodically. The full dataset could be stored in a database.

Up to you, Zoom. Can still be a fun project. Need more people to participate too.

I think the project has some potential. What could be done at least is automate the data collection from forum posts, by enforcing a strict way of posting results (so a parser could read it)

Then all that would be left to do is convince enough people to come post their results.

I see flaws in your research. I think you should write algorithms that randomly generate the stats that way you can run them 10 billion times over.

If you collect stats from players you will only generate the instances recreated in the game, instead of creating every possible instance in the game.

Edit: Keep an eye on my project in the off topic section. When its complete it will be of use to you.

https://docs.google.com/spreadsheet/ccc?key=0ArgrXJdE5Hd0dEdlU0lVemlwcTktUlVEM0Z0OFYxdXc&usp=sharing

i've been keeping track of my map drops for a while. haven't been logging xp gains but it may still be useful to you. 140ish maps logged.

"

Doobylee wrote:

I see flaws in your research. I think you should write algorithms that randomly generate the stats that way you can run them 10 billion times over.

If you collect stats from players you will only generate the instances recreated in the game, instead of creating every possible instance in the game.

Edit: Keep an eye on my project in the off topic section. When its complete it will be of use to you.

I'm confused about this suggestion. The purpose of this project is to study map drop rates and experience within maps in relation to it's map mods. How else could this data be gathered without players actually doing maps and reporting the results? On what data would you base a simulation?