The Ideal Log Parser

--------------------------------------------------------------------------------
Forward

Since I play a hunter this was written from a hunters perspective. I don't have the knowledge to discuss statistical ratios for other classes. I really don't know how to perform statistical analysis of DPS for mages, nor do I understand how to perform it for healers or tanks. I have some guesses, but I hope that by writing this, I can inspire those who do have that knowledge.


Introduction

WOW Webstats has been a huge help to WOW players in understanding raid performance. Combined with data from the armory, I have used it to develop tools by which I can measure and understand hunter DPS. Statistical Reference Guide - TKA Something

The time that it takes to manually compute all of the ratios in the linked post is quite time prohibitive. What I would like to see is WWS or something like it that automatically computes those ratios.

During the Burning Crusade good hunter DPS was critically dependent upon raid configuration. Hunters did not have any tools to provide hard data to raid leaders, we did not know how to measure buffing in a raid. After I identified a measure and performed a study, we had a tool to use. One hunter used statistical tools to address some concerns that he had about raid configuration.


“As I suspected much of the less than expected performance is due to buff issues (prot wars tank with devastate cycle so things like Sunder should be 100% present for most raids, though WWS reviews by me did find Faerie Fire there way less than 100% of the time, and no use of CoR by our warlocks). “

“The most interesting comment was the bottom 9% in terms of dps effectiveness in general from raid composition. That went down a treat at the officers meeting let me tell you.” http://forums.tkasomething.com/showt...t=12322&page=2


“So the maths led to one of the rogues actually being dropped from the coveted feral druid group for our raid last night, our first time in BT . . . . I actually think a penny dropped last night for a few folks to see people who've been held up as the 'supreme' dps'ers of the guild to be left behind by such a collosal margin simply by reallocating one buff from one person to another in the same group of 25.” http://forums.tkasomething.com/showt...t=12322&page=3
Raid configuration is just one of the many aspects of game play that we can use statistical analysis to improve. Being able to have a parser that automatically computes these ratios is a big help to all of us. No one would do all this work by hand.


Benefits

Here are some of the benefits that can be obtained from statistical analysis, along with some links to studies that I have performed in these areas.

Improve personal DPS
……. Grades for BM Hunters
…….Grades for BM Hunters - TKA Something
Understand shot rotations better
…….Data on the effectiveness of Shot Rotations
…….Data on the effectiveness of Shot Rotations - TKA Something
…….Data regarding the effectiveness of Beastmaster shot rotations
…….Data regarding the effectiveness of Beastmaster shot rotations - TKA Something
Optimize pet utilization
Measure the effectiveness of Macros on shot rotations
Class leaders can monitor performance
Separate hunter performance from raid organization
…….How to measure bad buffing and raid configuration
…….How to measure bad buffing and raid configuration - TKA Something
…….Data re: Buffs and raid config impact on hunter DPS
…….Data re: Buffs and raid config impact on hunter DPS - TKA Something
Enable raid leaders to measure the effectiveness of raid configurations
Measure the effect of latency + human reaction time and understand its effect on DPS


Known measurement tools

WOW Webstats Wow Web Stats
Wowmeter WoW Meter Online - The Best World of Warcraft Combat Log Analyse System!Fully support Wrath of the Lich King!
StatisCL stasiscl - Google Code
The ALP Project The ALP Project - TKA Something
Racy’s Log parser Wow Log Parser
Recap (Hawksy) Damage meter An error has occurred - Curse

Unfortunately none of these measurement tools use any data from the armory so they do not come close to realizing their potential, and by themselves are not very useful for statistical analysis. Very few of them compute ratios, and those that compute ratios do not compute many ratios. For example Recap computes the time per shot.

WOW Webstats is the best established of the log parsers, and contains great data. Since it is a centralized database it can also produce benchmarks. There is tremendous potential for improvement of the quality of WWS. However, I don’t think that the designers or the users are aware of the potential improvemnts.

I don’t know much about Wowmeter or StatisCL. There is a post at TKASomething that discusses them. WWS alternative? - TKA Something.

The ALP project appears to have been abandoned. It was attempting to accomplish most of the things that the ideal log parser would do.

Racy’s Log parser is available on curse.com. An error has occurred - Curse I tried downloading a long time ago and could not get it to work. I am not a programmer, I probably could get it to work if I spent the time to download the newest version of Java and knew more about Java. However, it does have some screen shots and it does not compute all of the ratios that I would like. I think that it does have potential.

Recap is the only damage meter that I have found that meets all of my criteria. Recount is close but not quite there. While it has great graphics, it does not provide the numerical detail that Recap does.

Ideal parser

Would be able to generate output in a variety of formats, the most important of which would be in plain text. That way a raid leader could print out the reports needed and post them on ANY guilds website. The program should be able to generate offline reports that are very detailed and would support queries by the operator. Program should also be able to generate a database that could be used to generate interactive reports like Wow Webstats does.

Would be able to compile statistical data from many fights and be able to use that as benchmarks in reports.

Would be able to accept data from the armory. It would probably need to be manually entered, since buffs are stored in the armory and would distort data.

Would be simple to use. The user would just have to download a program and run it. They would not have to learn any programming languages or download any other programs to make it run.

Would be open source and continually expanding. No one person can understand the whole scope of the project. (Maybe if they have no job but WOW, but even then, they would probably need a couple of assistants). Programmers could create add-ons for classes or other aspects.