Data mining method for League of Legends Stats

I have a lot of people asking me how I get my data for game statistics I am posting reports on. Hopefully this post answers most of the questions.

Skip this paragraph if you know how google works. As you know google and other search engines use applications capped crawlers that aimlessly scout the internet and download everything they can find usefully and store it for the search engine application to process. The data is then indexed by indexing applications and stored into a specialized database. When you search using a keyword google goes into the database and gets what you might be looking for.

How I get data is similar. Instead of indexing and looking for data aimlessly my tool indexes very specific pages and knows exactly what its looking for making it very efficient. Efficiency saves my computer as well as web server resources.

Where do I get data from?
Anywhere there is valuable data. Currently only lolbase and riot ranking ladders are indexed. Unfortunately riotstats.com is still under heavy development and has bugs and issues as well as missing a lot of valuable statistics.

I would recommend using JaxUploader from riotstats.com just because its uploading data to BOTH lolbase and riotstats.

Got more questions? Let me know.

Data scope and random factor

After looking at the data and some graphs I realized that there is substantial random variable that can decide a game. In order to get a crisp data large data sample is crucial. This especially is true for x2 champ combo.

I might release individual champ success ratio every few days while x2 champ combo every week or two. Effectively that will average data hopefully enough to remove most of the random factor and expose the interesting trends.

LoL Champion Combo Success Data [May 20-24]

Table below is sortable (click column title)

Here is table of Champ Success Ratios for both US and EU server in period between May 20-24, 2011. Keep in mind while looking at "Avg Player Elo" that since all Elo data coming from Riot Ranked ladders, players that are lower then 1200 Elo are represented by an Elo of 0 (zero).

There are 2,850 possible 2 champion combinations using 76 champs. These are 25 most successful.

With next set of data would be interesting to see how patch influenced the combos.

Champ1	Champ2	Avg Player Elo	Wins	Losses	1000*wins/losses	Win %
Maokai	Anivia	238.0	47	19	2473	71
Rumble	Fiddlesticks	207.0	61	29	2103	67
Twitch	Rumble	289.0	43	21	2047	67
Sona	Karthus	324.0	69	34	2029	66
Singed	Brand	309.0	112	62	1806	64
Warwick	Soraka	394.0	227	129	1759	63
Zilean	Nocturne	343.0	73	42	1738	63
Sona	Evelynn	207.0	116	67	1731	63
Heimerdinger	Caitlyn	258.0	129	75	1720	63
Soraka	Rumble	406.0	94	55	1709	63
Kassadin	Heimerdinger	130.0	41	24	1708	63
Irelia	Fiddlesticks	337.0	73	44	1659	62
Rumble	Rammus	325.0	176	107	1644	62
Sona	Corki	372.0	64	39	1641	62
Vladimir	Janna	431.0	263	162	1623	61
Swain	Rumble	305.0	60	37	1621	61
Urgot	Malphite	398.0	50	31	1612	61
Mordekaiser	Evelynn	307.0	256	159	1610	61
Mordekaiser	Maokai	214.0	80	50	1600	61
Shen	Rumble	347.0	145	91	1593	61
Soraka	Poppy	494.0	43	27	1592	61
Sona	Malzahar	367.0	155	99	1565	61
Udyr	Sona	292.0	72	46	1565	61
Vayne	Soraka	411.0	343	225	1524	60
Soraka	Lee Sin	465.0	59	39	1512	60

LoL Champion Success Data for May 20-24

This data is coming from lolbase.net and is generated from game logs uploaded to lolbase within past few days. There is also new data available: "Average Player Elo". Keep in mind while looking at "Avg Player Elo" that since all Elo data coming from Riot Ranked ladders, players that are lower then 1200 Elo are represented by an Elo of 0 (zero).

CLICK ON COLUMN TITLE TO SORT

Champ	Avg Player Elo	Wins	Losses	1000*wins/losses	Win %
Rumble	318.3	2001	1433	1396	58
Sona	285.0	1893	1370	1381	58
Soraka	402.6	1643	1261	1302	56
Janna	320.5	2087	1661	1256	55
Karthus	270.8	1084	873	1241	55
Amumu	345.3	3304	2723	1213	54
Mordekaiser	293.1	3397	2846	1193	54
Brand	279.4	2000	1761	1135	53
Malzahar	330.5	2464	2112	1166	53
Nocturne	286.2	2165	1910	1133	53
Vladimir	333.2	4083	3540	1153	53
Anivia	316.5	1077	973	1106	52
Evelynn	282.1	2383	2147	1109	52
Galio	300.3	421	379	1110	52
Irelia	394.6	2006	1799	1115	52
Singed	343.6	1677	1503	1115	52
Taric	382.2	798	720	1108	52
Twisted Fate	339.9	2691	2459	1094	52
Alistar	302.9	5313	5008	1060	51
Blitzcrank	354.1	1468	1376	1066	51
Caitlyn	289.2	2304	2173	1060	51
Fiddlesticks	198.4	1291	1197	1078	51
Heimerdinger	185.7	2054	1927	1065	51
Katarina	215.8	2645	2487	1063	51
Malphite	280.5	1734	1620	1070	51
Nunu	272.0	1913	1810	1056	51
Rammus	196.5	3467	3331	1040	51
Urgot	347.2	773	732	1056	51
Zilean	316.2	1080	1037	1041	51
Ashe	276.0	4695	4529	1036	50
Cassiopeia	288.1	869	863	1006	50
Corki	455.7	777	777	1000	50
Kennen	219.0	1872	1802	1038	50
Nasus	281.5	1538	1516	1014	50
Warwick	245.4	4534	4385	1033	50
Xin Zhao	219.3	7352	7101	1035	50
Annie	259.0	1680	1711	981	49
Gangplank	318.9	1561	1610	969	49
Jax	267.8	2773	2850	972	49
Lee Sin	404.4	1235	1257	982	49
Maokai	295.3	911	915	995	49
Miss Fortune	218.1	3703	3760	984	49
Poppy	286.3	1223	1257	972	49
Teemo	198.6	2997	3003	998	49
Trundle	398.6	713	727	980	49
Tryndamere	191.6	2542	2563	991	49
Jarvan IV	276.8	4364	4597	949	48
Kog'Maw	231.8	576	609	945	48
Swain	335.5	980	1060	924	48
Twitch	227.7	792	857	924	48
Vayne	287.5	6021	6513	924	48
Garen	271.6	1753	1948	899	47
Master Yi	176.3	2476	2724	908	47
Morgana	246.1	1186	1331	891	47
Renekton	341.7	1049	1142	918	47
Udyr	315.3	958	1077	889	47
Kassadin	341.9	902	1023	881	46
Lux	260.9	2124	2450	866	46
Nidalee	342.3	1482	1687	878	46
Ryze	268.6	1523	1766	862	46
Shaco	308.0	1707	1948	876	46
Shen	357.2	1908	2176	876	46
Sion	296.3	1065	1241	858	46
Cho'Gath	292.6	1349	1610	837	45
Dr. Mundo	224.5	1237	1476	838	45
Ezreal	352.2	1306	1573	830	45
Gragas	347.3	496	594	835	45
Kayle	300.2	668	812	822	45
Pantheon	253.1	1030	1218	845	45
Karma	342.2	482	608	792	44
Sivir	176.2	909	1139	798	44
Akali	252.8	1914	2466	776	43
LeBlanc	269.0	1080	1379	783	43
Olaf	299.7	1003	1281	782	43
Veigar	181.6	2743	3514	780	43
Tristana	231.2	882	1202	733	42

As you can see Rumble did really well I can't wait to see how well he will do after t nerf or the latest patch. Right after Rumble as always underrated support :)

Hello world :)

This blog is dedicated to my projects and personal findings in computing technology and programming.

As some of you may know, I have previously posted some statistics for a League of Legends (DotA clone) - I will continue doing so here.

I have to thank a lot of people for supporting me in my past projects such as LoLMasteriesTool
(http://www.leagueoflegends.com/board/showthread.php?p=4887295#post4887295).
I was surprises that so many people found it useful (about 25k downloads). I had pleasure to work with people from other continues and supporters to make it compatible with different language game clients and configurations as well as new and challenging features.

Recently, to my pleasant surprise, Riot (makers of League of Legends) release an official version of the tool embedded into a game
(http://www.leagueoflegends.com/news/multiple-mastery-pages-are-here). This creates an opportunity for me to dedicate more time for new projects. As you may guess currently I am digging for interesting game statistics that is uploaded to lolbase.net . Few next posts might have something to do with it. ;)

Note: As part of a bogging process I hope to also improve my writing skills. Constructional criticism is welcome for both technical and writing aspects of the posts :)

Thank you for your support.