{"id":1936,"date":"2019-08-14T00:38:51","date_gmt":"2019-08-13T15:38:51","guid":{"rendered":"http:\/\/141.164.34.82\/?p=1936"},"modified":"2019-08-14T00:40:44","modified_gmt":"2019-08-13T15:40:44","slug":"%eb%b3%80%ec%88%98-%ec%84%a0%ed%83%9d-%eb%b0%a9%eb%b2%95-%ec%84%a0%ed%98%95-%ed%9a%8c%ea%b7%80-%eb%b6%84%ec%84%9d%ec%9c%bc%eb%a1%9c-%eb%b3%80%ec%88%98%eb%a5%bc-%ec%84%a0%ed%83%9d%ed%95%9c%eb%8b%a4","status":"publish","type":"post","link":"http:\/\/ds.sumeun.org\/?p=1936","title":{"rendered":"\ubcc0\uc218 \uc120\ud0dd \ubc29\ubc95: \uc120\ud615 \ud68c\uadc0 \ubd84\uc11d\uc73c\ub85c \ubcc0\uc218\ub97c \uc120\ud0dd\ud55c\ub2e4\uba74?"},"content":{"rendered":"

\ubcc0\uc218 \uc120\ud0dd<\/h2>\n

\uc608\uce21 \ubcc0\uc218\uac00 \ub9ce\uc73c\uba74 \ubd84\uc11d\uc774 \uae4c\ub2e4\ub86d\ub2e4. \uadf8\ub9ac\uace0 \uc124\uba85\ubcc0\uc218\uac00 \uc801\uc744\uc218\ub85d \ubaa8\ud615\uc774 \uc880 \ub354 \uac04\uba85\ud558\ub2e4\ub294 \uc7a5\uc810\uc774 \uc788\ub2e4. <\/p>\n

iseed=6\r\n<\/code><\/pre>\n
\ubcc0\uc218 \uc120\ud0dd \ubc29\ubc95 : \uc120\ud615 \ud68c\uadc0 \ubd84\uc11d\uc5d0 \ub530\ub978 p-value?<\/h2>\n
\ucd5c\uadfc\uc5d0 \ubcc0\uc218\ub97c \uc120\ud0dd\uc73c\ub85c \uc120\ud615 \ubaa8\ud615\uc73c\ub85c \ub3cc\ub824\uc11c \uc720\uc758\ubbf8\ud55c \ubcc0\uc218\ub97c \ucd94\ub9b0 \ud6c4 \ub2e4\uc2dc \uc778\uacf5\uc2e0\uacbd\ub9dd\uc774\ub098 \ubd80\uc2a4\ud305 \uc54c\uace0\ub9ac\uc998\uacfc \uac19\uc740 \ube44\uc120\ud615 \ubaa8\ud615\uc73c\ub85c \ubd84\uc11d\ud558\uaca0\ub2e4\ub294 \uc598\uae30\ub97c \uc5ec\ub7ec \ubc88<\/strong> \ub4e4\uc5c8\ub2e4. \ud558\uc9c0\ub9cc \uc774 \ubc29\ubc95\uc740 \ube44\uc120\ud615\uc131<\/strong>, \uc0c1\ud638\uc791\uc6a9<\/strong>, \uadf8\ub9ac\uace0 \ub192\uc740 \uc124\uba85\ubcc0\uc218 \uc0c1\uad00<\/strong>\uc5d0 \ucde8\uc57d\ud558\ub2e4\ub294 \ub2e8\uc810\uc774 \uc788\ub2e4. <\/p>\n
\ub2e4\uc74c\uc758 \uc608\ub97c \ubcf4\uc790. dat<\/code>\uc5d0\ub294 \ubaa8\ub450 10\uac1c\uc758 \ubcc0\uc218\uac00 \uc788\ub2e4. <\/p>\n
\uc2e4\uc81c \ub370\uc774\ud130 \uc0dd\uc131 \ubaa8\ud615\uc740 \ub2e4\uc74c\uacfc \uac19\ub2e4. \uc774\ub54c \\(x_3\\) \uc740 \\(-1\\), \\(+1\\) \uac12\ub9cc \uac00\uc9c8 \uc218 \uc788\uace0, \ub098\uba38\uc9c0 \ubcc0\uc218\ub4e4\uc740 \ubaa8\ub450 \uc5f0\uc18d\ud615\uc774\ub2e4.<\/p>\n
\\[y = x_1^2 + x_2\\times x_3 + \\frac{1}{5}(x_4+x_5+x_6+x_7+x_8)+e, \\ \\ e\\sim\\mathcal{N}(0,2)\\]<\/p>\n
iseed=iseed+1\r\nset.seed(iseed)\r\n\r\nN = 1000\r\n\r\nsdErr = 1\r\n\r\nx1 <- runif(N, -3, 3)\r\nx11 <- x1^2 + rnorm(N, 0, 1)\r\n\r\nx2 <- runif(N, -3, 3)\r\nx3 <- sample(c(-1, 1), N, replace=TRUE)\r\n\r\nncolMatX <- 5\r\nncolMatX2 <- 2\r\n\r\nlibrary(mvtnorm)\r\nsigma <- matrix(\r\n  rep(0.97, ncolMatX*ncolMatX),\r\n  ncolMatX, ncolMatX)\r\ndiag(sigma) = 1\r\n\r\nmatX <- rmvnorm(N, mean=rep(0,ncolMatX), sigma=sigma)\r\nx40 <- apply(matX, 1, mean)\/ncolMatX\r\ncolnames(matX) <- paste0('x', 4:(ncolMatX+3))\r\n\r\nmatX2 <- rmvnorm(N, mean=rep(0,ncolMatX2), sigma=diag(ncolMatX2))\r\ncolnames(matX2) <- paste0('x', (ncolMatX+4):(ncolMatX+ncolMatX2+3))\r\n\r\ny <- x11 + x2*x3 + x40 + rnorm(N, 0, sdErr)\r\n\r\ndat <- data.frame(y, x1, x2, x3, matX, matX2)\r\n<\/code><\/pre>\n
dat<\/code>\uc5d0\ub294 \uacb0\uacfc \ubcc0\uc218 y<\/code>\uc640 \uc124\uba85\ubcc0\uc218 (\ud6c4\ubcf4) x1<\/code> ~ x10<\/code>\uc774 \uc800\uc7a5\ub418\uc5b4 \uc788\ub2e4. \uacb0\uacfc \ubcc0\uc218 y<\/code>\uc640 \uc124\uba85\ubcc0\uc218 x1<\/code> ~ x10<\/code>\uc758 \uad00\uacc4\ub294 \ub2e4\uc74c\uacfc \uac19\ub2e4.<\/p>\n
pairs(dat)\r\n<\/code><\/pre>\n
\"plot<\/p>\n
\uc774\uc5d0 \ub300\ud574 \uc120\ud615 \ud68c\uadc0 \ubd84\uc11d\uc744 \ud574\ubcf4\uba74, \ub300\ub2e4\uc218\uc758 \ubcc0\uc218\uac00 \uc720\uc758\ubbf8\ud558\uc9c0 \uc54a\ub2e4. \uadf8\ub9ac\uace0 \uad00\uacc4\uc5c6\ub294 \ubcc0\uc218\uac00 \ub9ce\uc744 \uc218\ub85d \uad00\uacc4\uc5c6\ub294 \ubcc0\uc218\uac00 \uc720\uc758\ubbf8\ud560 \ud655\ub960\ub3c4 \ucee4\uc9c4\ub2e4.<\/p>\n
fit <- lm(y ~ ., data=dat)\r\nsummary(fit)\r\n<\/code><\/pre>\n
## \r\n## Call:\r\n## lm(formula = y ~ ., data = dat)\r\n## \r\n## Residuals:\r\n##     Min      1Q  Median      3Q     Max \r\n## -7.7919 -2.7111 -0.3295  2.2054 12.1018 \r\n## \r\n## Coefficients:\r\n##             Estimate Std. Error t value Pr(>|t|)    \r\n## (Intercept)  2.88618    0.11244  25.668   <2e-16 ***\r\n## x1           0.06892    0.06578   1.048    0.295    \r\n## x2          -0.01823    0.06545  -0.279    0.781    \r\n## x3          -0.17481    0.11283  -1.549    0.122    \r\n## x4           0.49042    0.57470   0.853    0.394    \r\n## x5          -0.22494    0.57057  -0.394    0.693    \r\n## x6           0.51613    0.58882   0.877    0.381    \r\n## x7          -0.05237    0.58239  -0.090    0.928    \r\n## x8          -0.63142    0.57685  -1.095    0.274    \r\n## x9          -0.06280    0.11460  -0.548    0.584    \r\n## x10         -0.10342    0.10754  -0.962    0.336    \r\n## ---\r\n## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1\r\n## \r\n## Residual standard error: 3.553 on 989 degrees of freedom\r\n## Multiple R-squared:  0.007756,   Adjusted R-squared:  -0.002277 \r\n## F-statistic: 0.7731 on 10 and 989 DF,  p-value: 0.655\r\n<\/pre>\n
\uc65c\ub0d0\ud558\uba74 \uc120\ud615 \ubaa8\ud615\uc740 \ubd84\uc11d\uac00\uac00 \ube44\uc120\ud615 \ud568\uc218\ub97c \uc9c1\uc811 \ub123\uc5b4\uc8fc\uc9c0 \uc54a\ub294\ub2e4\uba74, \ube44\uc120\ud615\uc744 \uc7a1\uc544\ub0b4\uc9c0 \ubabb\ud558\uae30 \ub54c\ubb38\uc774\ub2e4.\uc0c1\ud638\uc791\uc6a9\uc758 \uacbd\uc6b0\ub3c4 \ub9c8\ucc2c\uac00\uc9c0\uc774\ub2e4. \uadf8\ub9ac\uace0 \uc124\uba85\ubcc0\uc218 \uac04\uc758 \ub192\uc740 \uc0c1\uad00\uad00\uacc4\ub294 \ub2e4\uc74c\uc758 VIF(variance inflation factor)\uc5d0\uc11c \ubcf4\ub4ef\uc774 \uacc4\uc218 \ucd94\uc815\ub7c9\uc758 \ubd84\uc0b0\uc744 \ub192\uc774\uace0, \uadf8\uc5d0 \ub530\ub77c \uacc4\uc218\uac00 \uc720\uc758\ubbf8\ud558\uc9c0 \uc54a\uc744 \uac00\ub2a5\uc131\uc774 \ub192\uc544\uc9c0\uae30 \ub54c\ubb38\uc774\ub2e4.<\/p>\n
library(car)\r\nvif(fit)\r\n<\/code><\/pre>\n
##        x1        x2        x3        x4        x5        x6        x7 \r\n##  1.014541  1.014495  1.008454 26.584158 26.365402 27.712375 26.790798 \r\n##        x8        x9       x10 \r\n## 26.482867  1.004744  1.005543\r\n<\/pre>\n
\ub300\uc548\uc740 \ub79c\ub364\ud3ec\uc2a4\ud2b8\uc640 \uac19\uc740 \ube44\uc120\ud615 \ubaa8\ud615\uc5d0\uc11c \ubcc0\uc218\uc911\uc694\ub3c4(variable importance)\ub97c \uc54c\uc544\ubcf4\ub294 \uac83\uc774\ub2e4. \ud2b9\ud788 \ub79c\ub364\ud3ec\ub808\uc2a4\ud2b8\ub294 \ubcc4\ub2e4\ub978 \uc0ac\uc804 \ucc98\ub9ac\uc5c6\uc774 \ubcc0\uc218\ub97c \uc54c\uc544\uc11c \uc120\ud0dd\ud574 \uc8fc\ub294 \ubaa8\ud615\uc73c\ub85c \uc720\uba85\ud558\ub2e4.<\/p>\n
library(randomForest)\r\n\r\nfitRF <- \r\n  randomForest(y ~ ., \r\n               data = dat, \r\n               importance = TRUE)\r\nvarImpPlot(fitRF)\r\n<\/code><\/pre>\n
\"plot<\/p>\n
\ub2e4\uc74c\uc758 \uc778\uacf5\uc2e0\uacbd\ub9dd\uacfc \ube44\uad50\ub97c \ud574\ubcf4\uc790. (\ubb3c\ub860 \ub79c\ub364\ud3ec\ub808\uc2a4\ud2b8\uc640 \uc778\uacf5\uc2e0\uacbd\ub9dd\uc740 \ud558\uc774\ud37c \ud30c\ub77c\uba54\ud130\uac00 \ub9ce\uc544\uc11c \ubb34\uc218\ud788 \ub9ce\uc740 \uac00\ub2a5\uc131\uc774 \uc874\uc7ac\ud568\uc744 \uc720\uc758\ud558\uc790.)<\/p>\n
library(nnet)\r\nfitnn <-\r\n  nnet(y ~ ., size = 20, maxit=1000, data=dat, linout=TRUE)\r\n<\/code><\/pre>\n
## # weights:  241\r\n## initial  value 19763.535175 \r\n## iter  10 value 8776.908610\r\n## iter  20 value 3787.234385\r\n## iter  30 value 3108.000264\r\n## iter  40 value 2534.895276\r\n## iter  50 value 2114.143262\r\n## iter  60 value 1935.900091\r\n## iter  70 value 1761.310014\r\n## iter  80 value 1654.889180\r\n## iter  90 value 1583.571300\r\n## iter 100 value 1524.088861\r\n## iter 110 value 1484.832976\r\n## iter 120 value 1458.524574\r\n## iter 130 value 1427.436051\r\n## iter 140 value 1404.922083\r\n## iter 150 value 1373.114164\r\n## iter 160 value 1342.454345\r\n## iter 170 value 1309.063005\r\n## iter 180 value 1275.425096\r\n## iter 190 value 1255.580611\r\n## iter 200 value 1241.547763\r\n## iter 210 value 1230.834946\r\n## iter 220 value 1221.474566\r\n## iter 230 value 1212.521158\r\n## iter 240 value 1204.189948\r\n## iter 250 value 1199.259770\r\n## iter 260 value 1196.770384\r\n## iter 270 value 1193.644864\r\n## iter 280 value 1192.297930\r\n## iter 290 value 1191.155442\r\n## iter 300 value 1190.676059\r\n## iter 310 value 1190.350277\r\n## iter 320 value 1189.934482\r\n## iter 330 value 1189.770896\r\n## iter 340 value 1189.637688\r\n## iter 350 value 1189.524756\r\n## iter 360 value 1189.228944\r\n## iter 370 value 1189.174962\r\n## iter 380 value 1189.143043\r\n## iter 390 value 1189.112372\r\n## iter 400 value 1189.084989\r\n## iter 410 value 1189.047733\r\n## iter 420 value 1189.010142\r\n## iter 430 value 1188.998243\r\n## iter 440 value 1188.984764\r\n## iter 450 value 1188.966988\r\n## iter 460 value 1188.959658\r\n## iter 470 value 1188.955405\r\n## iter 480 value 1188.952245\r\n## final  value 1188.951349 \r\n## converged\r\n<\/pre>\n
#mean((mean(y)-y)^2)\r\n#mean((predict(fitnn, newdata=dat[,-1])-y)^2)\r\n\r\nlibrary(NeuralNetTools)\r\ngarson(fitnn)\r\n<\/code><\/pre>\n
\"plot<\/p>\n
olden(fitnn)\r\n<\/code><\/pre>\n
\"plot<\/p>\n","protected":false},"excerpt":{"rendered":"
\ubcc0\uc218 \uc120\ud0dd \uc608\uce21 \ubcc0\uc218\uac00 \ub9ce\uc73c\uba74 \ubd84\uc11d\uc774 \uae4c\ub2e4\ub86d\ub2e4. \uadf8\ub9ac\uace0 \uc124\uba85\ubcc0\uc218\uac00 \uc801\uc744\uc218\ub85d \ubaa8\ud615\uc774 \uc880 \ub354 \uac04\uba85\ud558\ub2e4\ub294 \uc7a5\uc810\uc774 \uc788\ub2e4. iseed=6 \ubcc0\uc218 \uc120\ud0dd \ubc29\ubc95 : \uc120\ud615 \ud68c\uadc0 \ubd84\uc11d\uc5d0 \ub530\ub978 p-value? \ucd5c\uadfc\uc5d0 \ubcc0\uc218\ub97c \uc120\ud0dd\uc73c\ub85c \uc120\ud615 \ubaa8\ud615\uc73c\ub85c \ub3cc\ub824\uc11c \uc720\uc758\ubbf8\ud55c \ubcc0\uc218\ub97c \ucd94\ub9b0 \ud6c4 \ub2e4\uc2dc \uc778\uacf5\uc2e0\uacbd\ub9dd\uc774\ub098 \ubd80\uc2a4\ud305 \uc54c\uace0\ub9ac\uc998\uacfc \uac19\uc740 \ube44\uc120\ud615 \ubaa8\ud615\uc73c\ub85c \ubd84\uc11d\ud558\uaca0\ub2e4\ub294 \uc598\uae30\ub97c \uc5ec\ub7ec \ubc88 \ub4e4\uc5c8\ub2e4. \ud558\uc9c0\ub9cc \uc774 \ubc29\ubc95\uc740 \ube44\uc120\ud615\uc131, \uc0c1\ud638\uc791\uc6a9, \uadf8\ub9ac\uace0 \ub192\uc740 […]<\/p>\n","protected":false},"author":1,"featured_media":1940,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":[],"categories":[146,319],"tags":[370,368,367,366,321,338,369],"jetpack_featured_media_url":"http:\/\/ds.sumeun.org\/wp-content\/uploads\/2019\/08\/variableSelection01.png","_links":{"self":[{"href":"http:\/\/ds.sumeun.org\/index.php?rest_route=\/wp\/v2\/posts\/1936"}],"collection":[{"href":"http:\/\/ds.sumeun.org\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"http:\/\/ds.sumeun.org\/index.php?rest_route=\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"http:\/\/ds.sumeun.org\/index.php?rest_route=\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"http:\/\/ds.sumeun.org\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=1936"}],"version-history":[{"count":2,"href":"http:\/\/ds.sumeun.org\/index.php?rest_route=\/wp\/v2\/posts\/1936\/revisions"}],"predecessor-version":[{"id":1938,"href":"http:\/\/ds.sumeun.org\/index.php?rest_route=\/wp\/v2\/posts\/1936\/revisions\/1938"}],"wp:featuredmedia":[{"embeddable":true,"href":"http:\/\/ds.sumeun.org\/index.php?rest_route=\/wp\/v2\/media\/1940"}],"wp:attachment":[{"href":"http:\/\/ds.sumeun.org\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=1936"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"http:\/\/ds.sumeun.org\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=1936"},{"taxonomy":"post_tag","embeddable":true,"href":"http:\/\/ds.sumeun.org\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=1936"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}