Malkai Inos
Opus Arcana
Covert Intervention
492
   |
Posted - 2013.06.29 18:18:00 -
[1] - Quote
So much stuff
Choosing and tuning ones own personal fitting is a large part of Dust 154 and is considered by many as an important and unique aspect of th game. While the high amount of customization and the vast number of possible builds has earned praise from even the most critical reviews from the media it is also known that designing a potent, efficient and, most importantly, usable fit is always a challange and many players, especially some non EVE players, have trouble making sense of the amount of choice they are confronted with.
This is an attempt to design a search system that assists players, based on several interconnected factors, to find common and successfull suit fits and displays them in an intuitive and inspiring manner.
Dust 514 and spam prevention tech
The basic inspiration for this idea is based on some of the ways search engines like google and bing generate their results in a way that shows users documents (webpages and in case of dust: combinations of suits and modules i.e. fits) that are likely to be helpfull for the user. The goal is to create a fully dynamic assistant that allows players to fit a suit without having to dig through spreadsheets or read pages of guides that might already be outdated and thus counterproductive to the user's needs.
A requirement for this system to work is that the tranquility server needs to keep track of every single kill during every match, record the fit used aswell as the overall performance (kd/r, wp gain per life) in relation to its price to build a massive database of combat metrics. This database also needs to include all attributes of every item and be able to account for Character skills to optionally exclude combinations that are effective and common but fitting cost prohibitve for the searching player.
After hundreds of thousand of kills, statistical analysis of all the datapoints will show that certain combinations of suits, weapons and modules are more common than others and that usage of a particular gun in combination with a particular suit and certain modules yields a higher kd/r and/or wp gain per life on average than others. This results can be used to calculate statistical relations between those items which can be expressed in a dynamic ranking system and then displayed, based on this ranking system.
One Example:
A player could search for viable fits with the STD Gallente Assault rifle. The Dynamic Fitting Optimizer (working title, DFO for short) opens and searches the database for every fit that has ever been fielded while including this gun. The DFO would quickly see that STD Gallente Assault rifles are more often and more successfully used with Medium suits than with light or heavy suits as well as more likely to be combined with armor, shield or damage mods than with scan rez or melee mods thus already ranking some items higher than others.
In the center of the screen we see a smaller, stylized version of the fitting screen with all the module slots and the AR already occupying one light weapon slot. Selecting a slot will display the most highly ranked items in a layered fasion with similarly ranked modules residing next to each other and larger differences beeing expressed by the items occupying different layers going outwards the lower they're ranked by the system. Selecting any module will immediately trigger the DFO to recalculate all its rankings, based on now two specific items: The STD AR and the newly selected item.
This changes the ranking of all items in such a way that certain modules are now more inwards than before, some less and some will even no longer be displayed since they are no longer common and/or effective enough to be suggested the build at this stage or simply not possible to fit. Alternatively well ranked but unusable items could be still displayed but greyed or reded out.
Selecting any shield mod for example could exclude or move gallente suits further outwards since they are only rarely effective with shield shield mods. Picking the gallente suit first instead would make shield mods to no longer appear and move dmg mods for highs and armor mods for lows further inwards. Selecting three or four eq modules would exclude all non logi suits and change all other fitting suggestions accordingly.
With every decision made, the DFO ranking of every item changes every time, excluding more and more unusual or ineffective module combinations and highlighting the most used and effective modules in every intermediate step. Players can try and select different modules for every slot and observe how their choices affect the DFO suggestions. They can even start with an already complete fit then change the suit or other modules and watch the recommendations change to further improve the fitting.
Conclusion
Based on how much work is put into this system, players can choose to take skill requirements into account or to ignore them, wether or not suggestions get culled if they exceed the cpu/pg limit due to lacking core skills or wether they want only items to be shown that they already posess or all items available in the market. They can even decide wether they prioritize ISK cost or pure effectiveness for their fits by taking a cost benefit factor into the equation.
This system is based on realtime statistical ingame data and thus bound to be always up to date. It not only quickly catches particularly strong combinations to help players not to get lost in the inevitable FOTM race after every balance pass but also provides a comprehensive and easy to use dataset for the Devs to make informed balance choices and spot imbalances much more quickly than by manual examination of unprocessed data. It also gives players a tool to learn about sensible fitting choices by simply observing how recommendations change with every new decision and trying to get various usefull combinations. |