{"id":466,"date":"2018-09-11T22:12:45","date_gmt":"2018-09-11T13:12:45","guid":{"rendered":"https:\/\/best-biostatistics.com\/toukei-er\/entry\/random-forest-tuning-in-r\/"},"modified":"2024-10-12T22:43:44","modified_gmt":"2024-10-12T13:43:44","slug":"random-forest-tuning-in-r","status":"publish","type":"post","link":"https:\/\/best-biostatistics.com\/toukei-er\/entry\/random-forest-tuning-in-r\/","title":{"rendered":"R \u3067\u30e9\u30f3\u30c0\u30e0\u30d5\u30a9\u30ec\u30b9\u30c8\u3092\u6700\u9069\u5316\u3059\u308b\u65b9\u6cd5"},"content":{"rendered":"\n

\u30e9\u30f3\u30c0\u30e0\u30d5\u30a9\u30ec\u30b9\u30c8\u306f\u30c1\u30e5\u30fc\u30cb\u30f3\u30b0\u3057\u3066\u6700\u9069\u5316\u3059\u308b\u3002<\/p>\n\n\n\n

\u30c1\u30e5\u30fc\u30cb\u30f3\u30b0\u306f\u6c7a\u5b9a\u6728\u3092\u6700\u9069\u5316\u3059\u308b\u65b9\u6cd5\u3002<\/p>\n\n\n\n

\u30e9\u30f3\u30c0\u30e0\u30d5\u30a9\u30ec\u30b9\u30c8\u306e\u5834\u5408\u306f\u3001\u6c7a\u5b9a\u6728\u306e\u6570\u3068\u7279\u5fb4\u91cf\uff08\u8aac\u660e\u5909\u6570\uff09\u306e\u6570\u3092\u6700\u9069\u5316\u3059\u308b\u3002<\/p>\n\n\n\n\n\n\n\n

\u30e9\u30f3\u30c0\u30e0\u30d5\u30a9\u30ec\u30b9\u30c8\u306e\u30d1\u30c3\u30b1\u30fc\u30b8\u306e\u30a4\u30f3\u30b9\u30c8\u30fc\u30eb\u3068\u6e96\u5099<\/h2>\n\n\n\n

\u6700\u521d\u306b\u4e00\u56de\u3060\u3051\u30d1\u30c3\u30b1\u30fc\u30b8\u306e\u30a4\u30f3\u30b9\u30c8\u30fc\u30eb\u3092\u3059\u308b\u3002<\/p>\n\n\n\n

install.packages(\"randomForest\")<\/code><\/pre>\n\n\n\n
\u547c\u3073\u51fa\u3057\u3066\u4f7f\u3048\u308b\u3088\u3046\u306b\u6e96\u5099\u3059\u308b\u3002<\/p>\n\n\n\n
library(randomForest)<\/code><\/pre>\n\n\n\n
\u30e9\u30f3\u30c0\u30e0\u30d5\u30a9\u30ec\u30b9\u30c8\u306e\u6700\u9069\u5316\u306e\u305f\u3081\u306b\u4f8b\u984c\u3068\u3057\u3066\u4f7f\u7528\u3059\u308b\u30c7\u30fc\u30bf<\/h2>\n\n\n\n
MASS\u30d1\u30c3\u30b1\u30fc\u30b8\u306efgl\u30c7\u30fc\u30bf\u3092\u4f7f\u3046\u3002<\/p>\n\n\n\n
fgl\u306fForensic Glass Fragmens\u306e\u30c7\u30fc\u30bf\u3067\u3001\u6cd5\u533b\u5b66\u306b\u304a\u3051\u308b\u30ac\u30e9\u30b9\u7247\u306e\u30c7\u30fc\u30bf\u3002<\/p>\n\n\n\n
RI\u3068\u3044\u3046\u5c48\u6298\u7387Refractive Index\u30687\u7a2e\u985e\u306e\u91d1\u5c5e\u306e\u9178\u5316\u7269\u306e\u30c7\u30fc\u30bf\u3067\u3001\u30ac\u30e9\u30b9\u306e\u30bf\u30a4\u30d7\u3092\u8b58\u5225\u3059\u308b\u3068\u3044\u3046\u8ab2\u984c\u3002<\/p>\n\n\n\n
\u30ac\u30e9\u30b9\u306e\u30bf\u30a4\u30d7\u306f10\u5217\u76ee\u306e\u30c7\u30fc\u30bf\u3002<\/p>\n\n\n\n
library(MASS)\nhead(fgl)
table(fgl$type)<\/code><\/pre>\n\n\n\n
\u5148\u982d\u304b\u30895\u884c\u3092\u898b\u3066\u307f\u308b\u3068\u3053\u3093\u306a\u611f\u3058\u3002<\/p>\n\n\n\n
> head(fgl)\nRI    Na   Mg   Al    Si    K   Ca Ba   Fe type\n1  3.01 13.64 4.49 1.10 71.78 0.06 8.75  0 0.00 WinF\n2 -0.39 13.89 3.60 1.36 72.73 0.48 7.83  0 0.00 WinF\n3 -1.82 13.53 3.55 1.54 72.99 0.39 7.78  0 0.00 WinF\n4 -0.34 13.21 3.69 1.29 72.61 0.57 8.22  0 0.00 WinF\n5 -0.58 13.27 3.62 1.24 73.08 0.55 8.07  0 0.00 WinF\n6 -2.04 12.79 3.61 1.62 72.97 0.64 8.07  0 0.26 WinF\n\n# \u30ac\u30e9\u30b9\u306e\u30bf\u30a4\u30d7\u306f6\u7a2e\u985e\u3002\n > table(fgl$type)\nWinF WinNF   Veh   Con  Tabl  Head\n70    76    17    13     9    29\n<\/code><\/pre>\n\n\n\n
\u901a\u5e38\u306e\u30e9\u30f3\u30c0\u30e0\u30d5\u30a9\u30ec\u30b9\u30c8\u3092\u5b9f\u884c\u3057\u3066\u307f\u308b\u3002<\/p>\n\n\n\n
set.seed(2)\nfgl.rf <- randomForest(type ~ ., data=fgl)\nfgl.rf\nvarImpPlot(fgl.rf)<\/code><\/pre>\n\n\n\n
\u7d50\u679c\u306f\u4ee5\u4e0b\u306e\u901a\u308a\u3001\u8aac\u660e\u5909\u6570\u306f3\u3064\u3067\u3001\u30a8\u30e9\u30fc\u7387\u306f21.5\uff05<\/p>\n\n\n\n
randomForest()\u306e\u30c7\u30d5\u30a9\u30eb\u30c8\u306f\u5168\u8aac\u660e\u5909\u6570\u306e\u6570\uff089\u3064\uff09\u306e\u5e73\u65b9\u6839\u306a\u306e\u3067\u3001\u30d4\u30c3\u30bf\u30ea\u5408\u3063\u3066\u3044\u308b\u3002<\/p>\n\n\n\n
> fgl.rf\nCall:\nrandomForest(formula = type ~ ., data = fgl)\nType of random forest: classification\nNumber of trees: 500\nNo. of variables tried at each split: 3\nOOB estimate of  error rate: 21.5%\nConfusion matrix:\nWinF WinNF Veh Con Tabl Head class.error\nWinF    62     6   2   0    0    0   0.1142857\nWinNF   12    58   1   3    1    1   0.2368421\nVeh      7     3   7   0    0    0   0.5882353\nCon      0     3   0   9    0    1   0.3076923\nTabl     0     2   0   0    7    0   0.2222222\nHead     1     3   0   0    0   25   0.1379310\n<\/code><\/pre>\n\n\n\n
\u30b8\u30cb\u4e0d\u7d14\u5ea6\u306e\u6e1b\u5c11\u91cf\u3067\u898b\u308b\u91cd\u8981\u306a\u5909\u65703\u3064\u306f\u3001\u30de\u30b0\u30cd\u30b7\u30a6\u30e0\u3068\u30a2\u30eb\u30df\u30cb\u30a6\u30e0\u3068\u5c48\u6298\u7387\u3002<\/p>\n\n\n\n
\u30b8\u30cb\u4e0d\u7d14\u5ea6\u304c\u5927\u304d\u304f\u4e0b\u304c\u308b\u5909\u6570\u304c\u3001\u8b58\u5225\u306b\u5f79\u7acb\u3064\u91cd\u8981\u306a\u5909\u6570\u3002<\/p>\n\n\n\n
\"\"<\/figure>\n\n\n\n\n\n\n\n
\uff1e\uff1e\u3082\u3046\u7d71\u8a08\u3067\u60a9\u3080\u306e\u306f\u7d42\u308f\u308a\u306b\u3057\u307e\u305b\u3093\u304b\uff1f\u00a0<\/a><\/strong><\/span><\/p>\r\n\"\"<\/a>\r\n
\u21911\u4e07\u4eba\u4ee5\u4e0a\u306e\u533b\u7642\u5f93\u4e8b\u8005\u304c\u8cfc\u8aad\u4e2d<\/span><\/strong><\/span><\/p><\/div>
\u30e9\u30f3\u30c0\u30e0\u30d5\u30a9\u30ec\u30b9\u30c8\u3092\u6700\u9069\u5316\uff08\u30c1\u30e5\u30fc\u30cb\u30f3\u30b0\uff09\u3059\u308b<\/h2>\n\n\n\n
tuneRF()\u3092\u4f7f\u3046\u3002<\/p>\n\n\n\n
()\u5185\u306edoBest=T\u306f\u6700\u9069\u306a\u6c7a\u5b9a\u6728\u3092object\u3068\u3057\u3066\u8fd4\u3059\u30aa\u30d7\u30b7\u30e7\u30f3\u3002<\/p>\n\n\n\n
set.seed(2)\nfgl.res <- tuneRF(fgl[,-10], fgl[,10], doBest=T)\nfgl.res\nvarImpPlot(fgl.res)<\/code><\/pre>\n\n\n\n
\u7d50\u679c\u306f\u30a8\u30e9\u30fc\u738719.16\uff05\u306b\u306a\u3063\u305f\u3002<\/p>\n\n\n\n
\u8aac\u660e\u5909\u6570\u306f2\u3064\u306b\u6e1b\u3063\u305f\u3002<\/p>\n\n\n\n
> fgl.res <- tuneRF(fgl[,-10], fgl[,10], doBest=T)\nmtry = 3  OOB error = 22.43%\nSearching left ...\nmtry = 2        OOB error = 21.5%\n0.04166667 0.05\nSearching right ...\nmtry = 6        OOB error = 25.7%\n-0.1458333 0.05\n> fgl.res\nCall:\nrandomForest(x = x, y = y, mtry = res[which.min(res[, 2]), 1])\nType of random forest: classification\nNumber of trees: 500\nNo. of variables tried at each split: 2\nOOB estimate of  error rate: 19.16%\nConfusion matrix:\nWinF WinNF Veh Con Tabl Head class.error\nWinF    62     6   2   0    0    0   0.1142857\nWinNF    9    62   1   1    2    1   0.1842105\nVeh      7     4   6   0    0    0   0.6470588\nCon      0     2   0  10    0    1   0.2307692\nTabl     0     1   0   0    8    0   0.1111111\nHead     1     3   0   0    0   25   0.1379310\n<\/code><\/pre>\n\n\n\n
tuneRF()\u3067\u306f\u3001\u8aac\u660e\u5909\u6570\u304c\u3044\u304f\u3064\u306e\u6642\u306b\u30a8\u30e9\u30fc\u7387\u304c\u4e00\u756a\u5c0f\u3055\u304f\u306a\u308b\u304b\u304c\u81ea\u52d5\u7684\u306b\u56f3\u793a\u3055\u308c\u308b\u3002<\/p>\n\n\n\n
\"\"<\/figure>\n\n\n\n\n\n\n\n
\u30b8\u30cb\u4e0d\u7d14\u5ea6\u306e\u6e1b\u5c11\u91cf\u3067\u898b\u308b\u91cd\u8981\u306a\u5909\u6570\u4e0a\u4f4d 2 \u3064\u306f\u3001\u30de\u30b0\u30cd\u30b7\u30a6\u30e0\u3068\u30a2\u30eb\u30df\u30cb\u30a6\u30e0\u3002<\/p>\n\n\n\n
\"\"<\/figure>\n\n\n\n\n\n\n\n
\u30e9\u30f3\u30c0\u30e0\u30d5\u30a9\u30ec\u30b9\u30c8\u306e\u6700\u9069\u5316\u306e\u305f\u3081\u306b\u6c7a\u5b9a\u6728\u306e\u6570\u3092\u5909\u3048\u3066\u3084\u3063\u3066\u307f\u308b<\/h2>\n\n\n\n
\u6700\u5f8c\u306bntree\u3092\u30c7\u30d5\u30a9\u30eb\u30c8\u306e500\u304b\u30892000\u306b\u5897\u3084\u3057\u3066\u30a8\u30e9\u30fc\u7387\u306e\u5909\u5316\u3092\u898b\u3066\u307f\u308b\u3002<\/p>\n\n\n\n
set.seed(2)\nfgl.rf2 <- randomForest(type ~ ., data=fgl, mtry=2, ntree=2000)\nfgl.rf2\nplot(fgl.rf2)<\/code><\/pre>\n\n\n\n
\u6700\u7d42\u7684\u306a\u30a8\u30e9\u30fc\u7387\u306f19.63\uff05\u3068\u5927\u304d\u304f\u306f\u6539\u5584\u3057\u306a\u304b\u3063\u305f\u3002<\/p>\n\n\n\n
> fgl.rf2\nCall:\nrandomForest(formula = type ~ ., data = fgl, mtry = 2, ntree = 2000)\nType of random forest: classification\nNumber of trees: 2000\nNo. of variables tried at each split: 2\nOOB estimate of  error rate: 19.63%\nConfusion matrix:\nWinF WinNF Veh Con Tabl Head class.error\nWinF    62     6   2   0    0    0   0.1142857\nWinNF   10    62   1   1    1    1   0.1842105\nVeh      7     4   6   0    0    0   0.6470588\nCon      0     3   0   9    0    1   0.3076923\nTabl     0     1   0   0    8    0   0.1111111\nHead     1     3   0   0    0   25   0.1379310\n<\/code><\/pre>\n\n\n\n
\u9ed2\u306e\u5b9f\u7dda\u304c\u30a8\u30e9\u30fc\u7387\u3002<\/p>\n\n\n\n
\u305d\u308c\u4ee5\u5916\u306f\u30af\u30e9\u30b9\u30a8\u30e9\u30fc\u7387\u3002<\/p>\n\n\n\n
\u30c7\u30d5\u30a9\u30eb\u30c8\u306e500\u500b\u3067\u307b\u307c\u53ce\u675f\u3057\u3066\u3044\u308b\u306e\u304c\u898b\u3066\u53d6\u308c\u308b\u3002<\/p>\n\n\n\n
\"\"<\/figure>\n\n\n\n\n\n\n\n
\u307e\u3068\u3081<\/h2>\n\n\n\n
\u30e9\u30f3\u30c0\u30e0\u30d5\u30a9\u30ec\u30b9\u30c8\u3092\u30c1\u30e5\u30fc\u30cb\u30f3\u30b0\uff08\u6700\u9069\u5316\uff09\u3057\u3066\u307f\u305f\u3002<\/p>\n\n\n\n
\u7d71\u8a08\u30bd\u30d5\u30c8R\u306erandomForest\u30d1\u30c3\u30b1\u30fc\u30b8\u306b\u3042\u308btuneRF()\u3067\u5b9f\u884c\u3067\u304d\u308b\u3002<\/p>\n\n\n\n
\u3088\u308a\u30a8\u30e9\u30fc\u7387\u304c\u4f4e\u3044\u6c7a\u5b9a\u6728\u3092\u4f5c\u6210\u3059\u308b\u3053\u3068\u306b\u6210\u529f\u3057\u305f\u3002<\/p>\n","protected":false},"excerpt":{"rendered":"
\u30e9\u30f3\u30c0\u30e0\u30d5\u30a9\u30ec\u30b9\u30c8\u306f\u30c1\u30e5\u30fc\u30cb\u30f3\u30b0\u3057\u3066\u6700\u9069\u5316\u3059\u308b\u3002 \u30c1\u30e5\u30fc\u30cb\u30f3\u30b0\u306f\u6c7a\u5b9a\u6728\u3092\u6700\u9069\u5316\u3059\u308b\u65b9\u6cd5\u3002 \u30e9\u30f3\u30c0\u30e0\u30d5\u30a9\u30ec\u30b9\u30c8\u306e\u5834\u5408\u306f\u3001\u6c7a\u5b9a\u6728\u306e\u6570\u3068\u7279\u5fb4\u91cf\uff08\u8aac\u660e\u5909\u6570\uff09\u306e\u6570\u3092\u6700\u9069\u5316\u3059\u308b\u3002<\/p>\n","protected":false},"author":2,"featured_media":2744,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"swell_btn_cv_data":"","_jetpack_memberships_contains_paid_content":false,"footnotes":"","jetpack_publicize_message":"","jetpack_publicize_feature_enabled":true,"jetpack_social_post_already_shared":false,"jetpack_social_options":{"image_generator_settings":{"template":"highway","default_image_id":0,"font":"","enabled":false},"version":2}},"categories":[5,103],"tags":[],"class_list":["post-466","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-r","category-103"],"jetpack_publicize_connections":[],"jetpack_featured_media_url":"https:\/\/best-biostatistics.com\/toukei-er\/wp-content\/uploads\/2018\/09\/20180911220949.png","jetpack_sharing_enabled":true,"_links":{"self":[{"href":"https:\/\/best-biostatistics.com\/toukei-er\/wp-json\/wp\/v2\/posts\/466","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/best-biostatistics.com\/toukei-er\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/best-biostatistics.com\/toukei-er\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/best-biostatistics.com\/toukei-er\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/best-biostatistics.com\/toukei-er\/wp-json\/wp\/v2\/comments?post=466"}],"version-history":[{"count":2,"href":"https:\/\/best-biostatistics.com\/toukei-er\/wp-json\/wp\/v2\/posts\/466\/revisions"}],"predecessor-version":[{"id":2745,"href":"https:\/\/best-biostatistics.com\/toukei-er\/wp-json\/wp\/v2\/posts\/466\/revisions\/2745"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/best-biostatistics.com\/toukei-er\/wp-json\/wp\/v2\/media\/2744"}],"wp:attachment":[{"href":"https:\/\/best-biostatistics.com\/toukei-er\/wp-json\/wp\/v2\/media?parent=466"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/best-biostatistics.com\/toukei-er\/wp-json\/wp\/v2\/categories?post=466"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/best-biostatistics.com\/toukei-er\/wp-json\/wp\/v2\/tags?post=466"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}