Pages: [1] :: one page |
|
Author |
Thread Statistics | Show CCP posts - 0 post(s) |
krallistice
Not Connected
1
|
Posted - 2014.01.20 10:51:00 -
[1] - Quote
Hey,
i did some Data Mining with zKb on T1 Frigs to find out which Frigs are good against each other in Solo PvP. If you're interested you can the result in my Blog. I planned to do this bigger with more ships but unfortunately i don't have enough time right now :(.
http://kralli-pewpew.blogspot.de/p/statisticdata-mining-stuff.html
If you like, have any questions etc.. Feel free to comment or sent me a mail :)
(the rest of my Blog is in German, but if you understand it, feel free to read it :P ) |
fenistil
Ratchet Industries Dark Aether
103
|
Posted - 2014.01.20 11:21:00 -
[2] - Quote
That is some excellent stuff!
Just for clarification via an example. The total number of engagements you examind (let's say for Tristan vs. Rifter) is the sum of the "vs" numbers right? 259 vs. 23 totaling in 282 engagements out of which the tristan won 259.
Is that correct? Also would it not make any sense that you would display these numbers in %? Say if you fly a tristan against a rifter you have a 91% chance of winning over having to say: Out of 282 engagemenst I'll win 259.
It would be nice to add some aggregation of the data to the page: - how many killmails you included in the graph (after exclusions) - which ships were flown the most; relative to the total (Something like: Tristan 14% - 282 engagements)
Something like this would also be nice for the examined ships: - if you fly a tristan your favorite pray is: 3 most successfully killed ships (highest % of success not most kills in raw numbers) - and if you fly a tristan you are most likely be killed by: 3 ships that has the highest chance to kill you
Keep it up though! That's awesome... . |
krallistice
Not Connected
1
|
Posted - 2014.01.20 13:21:00 -
[3] - Quote
The Number are how often they killed their counterpart. So you are right.
Yeah, i had the % as edge labels, but for some sigma.js had problems with these, so i removed them :P
Yeah next time i will add stuff like most likely to be killed etc. You can find this out by yourself, if you campare the numbers, but it would be much nicer to have these direct.
I will post or add to the post the total numbers, etc when i get home. But it wanst that much in total, i was quite suprised how less solo killmails exist (maybe this number was a bit incorrect due to zKB definition of solo). |
fenistil
Ratchet Industries Dark Aether
103
|
Posted - 2014.01.20 13:54:00 -
[4] - Quote
I hear rumors about CCP serving kills themselves soon(tm). Maybe that will move the zKill 40-60% kill coverage closer to 100% giving you more accurate data.
One way to circumvent zKill's interesting definition of solo is to get _all_ the kills and process them yourself. With that I could probably help you and just provide you with the data in whatever format you need it. . |
God's Apples
AQUILA INC Verge of Collapse
303
|
Posted - 2014.01.20 17:12:00 -
[5] - Quote
Am I misinterpreting the graph or is the incursus undefeated?
The results make sense though. The tristan, breacher, and tormentor are able to project anywhere in scram range as well as having decent tanks and full tackle which means that they can't be hard countered by anything that isn't an MWD kiter (or OH'd web kiters like long point AB rail atron). |
krallistice
Not Connected
2
|
Posted - 2014.01.20 17:48:00 -
[6] - Quote
fenistil wrote:I hear rumors about CCP serving kills themselves soon(tm). Maybe that will move the zKill 40-60% kill coverage closer to 100% giving you more accurate data.
One way to circumvent zKill's interesting definition of solo is to get _all_ the kills and process them yourself. With that I could probably help you and just provide you with the data in whatever format you need it.
Here are the Numbers: (this is after i removed empty KMs (i dont have that data anymore), before i manually removed edges)
Crucifier Kills: 83.0 Killed: 97.0 Sum: 180.0 Executioner Kills: 408.0 Killed: 603.0 Sum: 1011.0 Punisher Kills: 128.0 Killed: 490.0 Sum: 618.0 Tormentor Kills: 646.0 Killed: 444.0 Sum: 1090.0 Condor Kills: 1386.0 Killed: 1570.0 Sum: 2956.0 Griffin Kills: 23.0 Killed: 138.0 Sum: 161.0 Kestrel Kills: 895.0 Killed: 1200.0 Sum: 2095.0 Merlin Kills: 1264.0 Killed: 1521.0 Sum: 2785.0 Atron Kills: 1640.0 Killed: 1719.0 Sum: 3359.0 Incursus Kills: 2519.0 Killed: 1442.0 Sum: 3961.0 Maulus Kills: 93.0 Killed: 164.0 Sum: 257.0 Tristan Kills: 2703.0 Killed: 1446.0 Sum: 4149.0 Breacher Kills: 1120.0 Killed: 478.0 Sum: 1598.0 Rifter Kills: 504.0 Killed: 1593.0 Sum: 2097.0 Slasher Kills: 705.0 Killed: 1192.0 Sum: 1897.0 Vigil Kills: 3.0 Killed: 23.0 Sum: 26.0
Total KM's: 14120.0
Yeah, i could get all Kills and then manually sort them out, but i didnt want to hammer their API. It seems to they already have big problems (i got many server side errors, up to 70%, while fetching the kills) to keep their services alive :P
God's Apples wrote:Am I misinterpreting the graph or is the incursus undefeated?
The results make sense though. The tristan, breacher, and tormentor are able to project anywhere in scram range as well as having decent tanks and full tackle which means that they can't be hard countered by anything that isn't an MWD kiter (or OH'd web kiters like long point AB rail atron).
Yes in this Graph the Incursus is undefeated (= Wins in most cases against their targets), although in the original graph the breacher and the tristan defeated the Incursus. I dropped these edges since they aren't very representative (having 11 more kills isnt really conclusive)
|
|
|
|
Pages: [1] :: one page |
First page | Previous page | Next page | Last page |