Работа с пакетами CARET и BORUTA на языке R
Elena Fedoseeva
2026-03-07
Задание 1
Установить пакет CARET, выполнить команду names(getModelInfo()), ознакомиться со списком доступных методов выбора признаков. Выполните графический разведочный анализ данных с использование функции featurePlot() для набора данных из справочного файла пакета CARET: x <- matrix(rnorm(505),ncol=5) y <- factor(rep(c(“A”, “B”), 25)) Сохранить полученные графики в .jpg файлы. Сделать выводы.
library(caret)## Loading required package: ggplot2## Loading required package: latticenames(getModelInfo())## [1] "ada" "AdaBag" "AdaBoost.M1"
## [4] "adaboost" "amdai" "ANFIS"
## [7] "avNNet" "awnb" "awtan"
## [10] "bag" "bagEarth" "bagEarthGCV"
## [13] "bagFDA" "bagFDAGCV" "bam"
## [16] "bartMachine" "bayesglm" "binda"
## [19] "blackboost" "blasso" "blassoAveraged"
## [22] "bridge" "brnn" "BstLm"
## [25] "bstSm" "bstTree" "C5.0"
## [28] "C5.0Cost" "C5.0Rules" "C5.0Tree"
## [31] "cforest" "chaid" "CSimca"
## [34] "ctree" "ctree2" "cubist"
## [37] "dda" "deepboost" "DENFIS"
## [40] "dnn" "dwdLinear" "dwdPoly"
## [43] "dwdRadial" "earth" "elm"
## [46] "enet" "evtree" "extraTrees"
## [49] "fda" "FH.GBML" "FIR.DM"
## [52] "foba" "FRBCS.CHI" "FRBCS.W"
## [55] "FS.HGD" "gam" "gamboost"
## [58] "gamLoess" "gamSpline" "gaussprLinear"
## [61] "gaussprPoly" "gaussprRadial" "gbm_h2o"
## [64] "gbm" "gcvEarth" "GFS.FR.MOGUL"
## [67] "GFS.LT.RS" "GFS.THRIFT" "glm.nb"
## [70] "glm" "glmboost" "glmnet_h2o"
## [73] "glmnet" "glmStepAIC" "gpls"
## [76] "hda" "hdda" "hdrda"
## [79] "HYFIS" "icr" "J48"
## [82] "JRip" "kernelpls" "kknn"
## [85] "knn" "krlsPoly" "krlsRadial"
## [88] "lars" "lars2" "lasso"
## [91] "lda" "lda2" "leapBackward"
## [94] "leapForward" "leapSeq" "Linda"
## [97] "lm" "lmStepAIC" "LMT"
## [100] "loclda" "logicBag" "LogitBoost"
## [103] "logreg" "lssvmLinear" "lssvmPoly"
## [106] "lssvmRadial" "lvq" "M5"
## [109] "M5Rules" "manb" "mda"
## [112] "Mlda" "mlp" "mlpKerasDecay"
## [115] "mlpKerasDecayCost" "mlpKerasDropout" "mlpKerasDropoutCost"
## [118] "mlpML" "mlpSGD" "mlpWeightDecay"
## [121] "mlpWeightDecayML" "monmlp" "msaenet"
## [124] "multinom" "mxnet" "mxnetAdam"
## [127] "naive_bayes" "nb" "nbDiscrete"
## [130] "nbSearch" "neuralnet" "nnet"
## [133] "nnls" "nodeHarvest" "null"
## [136] "OneR" "ordinalNet" "ordinalRF"
## [139] "ORFlog" "ORFpls" "ORFridge"
## [142] "ORFsvm" "ownn" "pam"
## [145] "parRF" "PART" "partDSA"
## [148] "pcaNNet" "pcr" "pda"
## [151] "pda2" "penalized" "PenalizedLDA"
## [154] "plr" "pls" "plsRglm"
## [157] "polr" "ppr" "pre"
## [160] "PRIM" "protoclass" "qda"
## [163] "QdaCov" "qrf" "qrnn"
## [166] "randomGLM" "ranger" "rbf"
## [169] "rbfDDA" "Rborist" "rda"
## [172] "regLogistic" "relaxo" "rf"
## [175] "rFerns" "RFlda" "rfRules"
## [178] "ridge" "rlda" "rlm"
## [181] "rmda" "rocc" "rotationForest"
## [184] "rotationForestCp" "rpart" "rpart1SE"
## [187] "rpart2" "rpartCost" "rpartScore"
## [190] "rqlasso" "rqnc" "RRF"
## [193] "RRFglobal" "rrlda" "RSimca"
## [196] "rvmLinear" "rvmPoly" "rvmRadial"
## [199] "SBC" "sda" "sdwd"
## [202] "simpls" "SLAVE" "slda"
## [205] "smda" "snn" "sparseLDA"
## [208] "spikeslab" "spls" "stepLDA"
## [211] "stepQDA" "superpc" "svmBoundrangeString"
## [214] "svmExpoString" "svmLinear" "svmLinear2"
## [217] "svmLinear3" "svmLinearWeights" "svmLinearWeights2"
## [220] "svmPoly" "svmRadial" "svmRadialCost"
## [223] "svmRadialSigma" "svmRadialWeights" "svmSpectrumString"
## [226] "tan" "tanSearch" "treebag"
## [229] "vbmpRadial" "vglmAdjCat" "vglmContRatio"
## [232] "vglmCumulative" "widekernelpls" "WM"
## [235] "wsrf" "xgbDART" "xgbLinear"
## [238] "xgbTree" "xyf"x <- matrix(rnorm(50*5),ncol=5)
y <- factor(rep(c("A", "B"), 25))
featurePlot(x, y)
featurePlot(x, y, plot = "pairs") # 
featurePlot(x, y, plot = "density") # 
jpeg("featurePlot.jpg")
featurePlot(x, y)
jpeg("featurePlot_pairs.jpg")
featurePlot(x, y, plot = "pairs")
jpeg("featurePlot_density.jpg")
featurePlot(x, y, plot = "density")“pairs”: Визуально точки классов A и B полностью перемешаны. Нет выраженных кластеров или линейных зависимостей.
“density”: Кривые плотности для классов A и B практически накладываются друг на друга для всех пяти признаков. Это говорит о том, что средние значения и дисперсии признаков в обоих классах идентичны.
Исходя из графиков, данные признаки не обладают предсказательной силой.
Задание 2
С использование функций из пакета Fselector [2] определить важность признаков для решения задачи классификации. Использовать набор data(iris). Сделать выводы.
library(FSelector)
data(iris)
head(iris)## Sepal.Length Sepal.Width Petal.Length Petal.Width Species
## 1 5.1 3.5 1.4 0.2 setosa
## 2 4.9 3.0 1.4 0.2 setosa
## 3 4.7 3.2 1.3 0.2 setosa
## 4 4.6 3.1 1.5 0.2 setosa
## 5 5.0 3.6 1.4 0.2 setosa
## 6 5.4 3.9 1.7 0.4 setosaПрименяется функция information.gain() для вычисления важности признаков. Этот метод оценивает, насколько много информации о целевом классе (Species) дает каждый признак.
inf <- information.gain(Species ~ ., data=iris)
inf## attr_importance
## Sepal.Length 0.4521286
## Sepal.Width 0.2672750
## Petal.Length 0.9402853
## Petal.Width 0.9554360Задание 3
С использованием функции discretize() из пакета arules выполните преобразование непрерывной переменной в категориальную [3] различными методами: «interval» (равная ширина интервала), «frequency» (равная частота), «cluster» (кластеризация) и «fixed» (категории задают границы интервалов). Используйте набор данных iris. Сделайте выводы
library(arules)## Loading required package: Matrix##
## Attaching package: 'arules'## The following objects are masked from 'package:base':
##
## abbreviate, writedata(iris)
iris_interval <- discretize(iris$Sepal.Width, method="interval", breaks=3)
print(iris_interval) # equal interval width## [1] [2.8,3.6) [2.8,3.6) [2.8,3.6) [2.8,3.6) [2.8,3.6) [3.6,4.4] [2.8,3.6)
## [8] [2.8,3.6) [2.8,3.6) [2.8,3.6) [3.6,4.4] [2.8,3.6) [2.8,3.6) [2.8,3.6)
## [15] [3.6,4.4] [3.6,4.4] [3.6,4.4] [2.8,3.6) [3.6,4.4] [3.6,4.4] [2.8,3.6)
## [22] [3.6,4.4] [2.8,3.6) [2.8,3.6) [2.8,3.6) [2.8,3.6) [2.8,3.6) [2.8,3.6)
## [29] [2.8,3.6) [2.8,3.6) [2.8,3.6) [2.8,3.6) [3.6,4.4] [3.6,4.4] [2.8,3.6)
## [36] [2.8,3.6) [2.8,3.6) [2.8,3.6) [2.8,3.6) [2.8,3.6) [2.8,3.6) [2,2.8)
## [43] [2.8,3.6) [2.8,3.6) [3.6,4.4] [2.8,3.6) [3.6,4.4] [2.8,3.6) [3.6,4.4]
## [50] [2.8,3.6) [2.8,3.6) [2.8,3.6) [2.8,3.6) [2,2.8) [2,2.8) [2,2.8)
## [57] [2.8,3.6) [2,2.8) [2.8,3.6) [2,2.8) [2,2.8) [2.8,3.6) [2,2.8)
## [64] [2.8,3.6) [2.8,3.6) [2.8,3.6) [2.8,3.6) [2,2.8) [2,2.8) [2,2.8)
## [71] [2.8,3.6) [2,2.8) [2,2.8) [2,2.8) [2.8,3.6) [2.8,3.6) [2,2.8)
## [78] [2.8,3.6) [2.8,3.6) [2,2.8) [2,2.8) [2,2.8) [2,2.8) [2,2.8)
## [85] [2.8,3.6) [2.8,3.6) [2.8,3.6) [2,2.8) [2.8,3.6) [2,2.8) [2,2.8)
## [92] [2.8,3.6) [2,2.8) [2,2.8) [2,2.8) [2.8,3.6) [2.8,3.6) [2.8,3.6)
## [99] [2,2.8) [2,2.8) [2.8,3.6) [2,2.8) [2.8,3.6) [2.8,3.6) [2.8,3.6)
## [106] [2.8,3.6) [2,2.8) [2.8,3.6) [2,2.8) [2.8,3.6) [2.8,3.6) [2,2.8)
## [113] [2.8,3.6) [2,2.8) [2,2.8) [2.8,3.6) [2.8,3.6) [3.6,4.4] [2,2.8)
## [120] [2,2.8) [2.8,3.6) [2,2.8) [2,2.8) [2,2.8) [2.8,3.6) [2.8,3.6)
## [127] [2,2.8) [2.8,3.6) [2,2.8) [2.8,3.6) [2,2.8) [3.6,4.4] [2,2.8)
## [134] [2,2.8) [2,2.8) [2.8,3.6) [2.8,3.6) [2.8,3.6) [2.8,3.6) [2.8,3.6)
## [141] [2.8,3.6) [2.8,3.6) [2,2.8) [2.8,3.6) [2.8,3.6) [2.8,3.6) [2,2.8)
## [148] [2.8,3.6) [2.8,3.6) [2.8,3.6)
## attr(,"discretized:breaks")
## [1] 2.0 2.8 3.6 4.4
## attr(,"discretized:method")
## [1] interval
## Levels: [2,2.8) [2.8,3.6) [3.6,4.4]iris_frequency <- discretize(iris$Sepal.Width, method="frequency", breaks=3)
print(iris_frequency) # equal frequency## [1] [3.2,4.4] [2.9,3.2) [3.2,4.4] [2.9,3.2) [3.2,4.4] [3.2,4.4] [3.2,4.4]
## [8] [3.2,4.4] [2.9,3.2) [2.9,3.2) [3.2,4.4] [3.2,4.4] [2.9,3.2) [2.9,3.2)
## [15] [3.2,4.4] [3.2,4.4] [3.2,4.4] [3.2,4.4] [3.2,4.4] [3.2,4.4] [3.2,4.4]
## [22] [3.2,4.4] [3.2,4.4] [3.2,4.4] [3.2,4.4] [2.9,3.2) [3.2,4.4] [3.2,4.4]
## [29] [3.2,4.4] [3.2,4.4] [2.9,3.2) [3.2,4.4] [3.2,4.4] [3.2,4.4] [2.9,3.2)
## [36] [3.2,4.4] [3.2,4.4] [3.2,4.4] [2.9,3.2) [3.2,4.4] [3.2,4.4] [2,2.9)
## [43] [3.2,4.4] [3.2,4.4] [3.2,4.4] [2.9,3.2) [3.2,4.4] [3.2,4.4] [3.2,4.4]
## [50] [3.2,4.4] [3.2,4.4] [3.2,4.4] [2.9,3.2) [2,2.9) [2,2.9) [2,2.9)
## [57] [3.2,4.4] [2,2.9) [2.9,3.2) [2,2.9) [2,2.9) [2.9,3.2) [2,2.9)
## [64] [2.9,3.2) [2.9,3.2) [2.9,3.2) [2.9,3.2) [2,2.9) [2,2.9) [2,2.9)
## [71] [3.2,4.4] [2,2.9) [2,2.9) [2,2.9) [2.9,3.2) [2.9,3.2) [2,2.9)
## [78] [2.9,3.2) [2.9,3.2) [2,2.9) [2,2.9) [2,2.9) [2,2.9) [2,2.9)
## [85] [2.9,3.2) [3.2,4.4] [2.9,3.2) [2,2.9) [2.9,3.2) [2,2.9) [2,2.9)
## [92] [2.9,3.2) [2,2.9) [2,2.9) [2,2.9) [2.9,3.2) [2.9,3.2) [2.9,3.2)
## [99] [2,2.9) [2,2.9) [3.2,4.4] [2,2.9) [2.9,3.2) [2.9,3.2) [2.9,3.2)
## [106] [2.9,3.2) [2,2.9) [2.9,3.2) [2,2.9) [3.2,4.4] [3.2,4.4] [2,2.9)
## [113] [2.9,3.2) [2,2.9) [2,2.9) [3.2,4.4] [2.9,3.2) [3.2,4.4] [2,2.9)
## [120] [2,2.9) [3.2,4.4] [2,2.9) [2,2.9) [2,2.9) [3.2,4.4] [3.2,4.4]
## [127] [2,2.9) [2.9,3.2) [2,2.9) [2.9,3.2) [2,2.9) [3.2,4.4] [2,2.9)
## [134] [2,2.9) [2,2.9) [2.9,3.2) [3.2,4.4] [2.9,3.2) [2.9,3.2) [2.9,3.2)
## [141] [2.9,3.2) [2.9,3.2) [2,2.9) [3.2,4.4] [3.2,4.4] [2.9,3.2) [2,2.9)
## [148] [2.9,3.2) [3.2,4.4] [2.9,3.2)
## attr(,"discretized:breaks")
## [1] 2.0 2.9 3.2 4.4
## attr(,"discretized:method")
## [1] frequency
## Levels: [2,2.9) [2.9,3.2) [3.2,4.4]iris_cluster <- discretize(iris$Sepal.Width, method="cluster", breaks=3)
print(iris_cluster) # k-means clustering## [1] [3.28,4.4] [2.69,3.28) [2.69,3.28) [2.69,3.28) [3.28,4.4] [3.28,4.4]
## [7] [3.28,4.4] [3.28,4.4] [2.69,3.28) [2.69,3.28) [3.28,4.4] [3.28,4.4]
## [13] [2.69,3.28) [2.69,3.28) [3.28,4.4] [3.28,4.4] [3.28,4.4] [3.28,4.4]
## [19] [3.28,4.4] [3.28,4.4] [3.28,4.4] [3.28,4.4] [3.28,4.4] [3.28,4.4]
## [25] [3.28,4.4] [2.69,3.28) [3.28,4.4] [3.28,4.4] [3.28,4.4] [2.69,3.28)
## [31] [2.69,3.28) [3.28,4.4] [3.28,4.4] [3.28,4.4] [2.69,3.28) [2.69,3.28)
## [37] [3.28,4.4] [3.28,4.4] [2.69,3.28) [3.28,4.4] [3.28,4.4] [2,2.69)
## [43] [2.69,3.28) [3.28,4.4] [3.28,4.4] [2.69,3.28) [3.28,4.4] [2.69,3.28)
## [49] [3.28,4.4] [3.28,4.4] [2.69,3.28) [2.69,3.28) [2.69,3.28) [2,2.69)
## [55] [2.69,3.28) [2.69,3.28) [3.28,4.4] [2,2.69) [2.69,3.28) [2.69,3.28)
## [61] [2,2.69) [2.69,3.28) [2,2.69) [2.69,3.28) [2.69,3.28) [2.69,3.28)
## [67] [2.69,3.28) [2.69,3.28) [2,2.69) [2,2.69) [2.69,3.28) [2.69,3.28)
## [73] [2,2.69) [2.69,3.28) [2.69,3.28) [2.69,3.28) [2.69,3.28) [2.69,3.28)
## [79] [2.69,3.28) [2,2.69) [2,2.69) [2,2.69) [2.69,3.28) [2.69,3.28)
## [85] [2.69,3.28) [3.28,4.4] [2.69,3.28) [2,2.69) [2.69,3.28) [2,2.69)
## [91] [2,2.69) [2.69,3.28) [2,2.69) [2,2.69) [2.69,3.28) [2.69,3.28)
## [97] [2.69,3.28) [2.69,3.28) [2,2.69) [2.69,3.28) [3.28,4.4] [2.69,3.28)
## [103] [2.69,3.28) [2.69,3.28) [2.69,3.28) [2.69,3.28) [2,2.69) [2.69,3.28)
## [109] [2,2.69) [3.28,4.4] [2.69,3.28) [2.69,3.28) [2.69,3.28) [2,2.69)
## [115] [2.69,3.28) [2.69,3.28) [2.69,3.28) [3.28,4.4] [2,2.69) [2,2.69)
## [121] [2.69,3.28) [2.69,3.28) [2.69,3.28) [2.69,3.28) [3.28,4.4] [2.69,3.28)
## [127] [2.69,3.28) [2.69,3.28) [2.69,3.28) [2.69,3.28) [2.69,3.28) [3.28,4.4]
## [133] [2.69,3.28) [2.69,3.28) [2,2.69) [2.69,3.28) [3.28,4.4] [2.69,3.28)
## [139] [2.69,3.28) [2.69,3.28) [2.69,3.28) [2.69,3.28) [2.69,3.28) [2.69,3.28)
## [145] [3.28,4.4] [2.69,3.28) [2,2.69) [2.69,3.28) [3.28,4.4] [2.69,3.28)
## attr(,"discretized:breaks")
## [1] 2.000000 2.691466 3.278481 4.400000
## attr(,"discretized:method")
## [1] cluster
## Levels: [2,2.69) [2.69,3.28) [3.28,4.4]iris_fixed <- discretize(iris$Sepal.Width, method = "fixed", breaks = c(-Inf, 2.5, 3.5, Inf))
print(iris_fixed)## [1] [3.5, Inf] [2.5,3.5) [2.5,3.5) [2.5,3.5) [3.5, Inf] [3.5, Inf]
## [7] [2.5,3.5) [2.5,3.5) [2.5,3.5) [2.5,3.5) [3.5, Inf] [2.5,3.5)
## [13] [2.5,3.5) [2.5,3.5) [3.5, Inf] [3.5, Inf] [3.5, Inf] [3.5, Inf]
## [19] [3.5, Inf] [3.5, Inf] [2.5,3.5) [3.5, Inf] [3.5, Inf] [2.5,3.5)
## [25] [2.5,3.5) [2.5,3.5) [2.5,3.5) [3.5, Inf] [2.5,3.5) [2.5,3.5)
## [31] [2.5,3.5) [2.5,3.5) [3.5, Inf] [3.5, Inf] [2.5,3.5) [2.5,3.5)
## [37] [3.5, Inf] [3.5, Inf] [2.5,3.5) [2.5,3.5) [3.5, Inf] [-Inf,2.5)
## [43] [2.5,3.5) [3.5, Inf] [3.5, Inf] [2.5,3.5) [3.5, Inf] [2.5,3.5)
## [49] [3.5, Inf] [2.5,3.5) [2.5,3.5) [2.5,3.5) [2.5,3.5) [-Inf,2.5)
## [55] [2.5,3.5) [2.5,3.5) [2.5,3.5) [-Inf,2.5) [2.5,3.5) [2.5,3.5)
## [61] [-Inf,2.5) [2.5,3.5) [-Inf,2.5) [2.5,3.5) [2.5,3.5) [2.5,3.5)
## [67] [2.5,3.5) [2.5,3.5) [-Inf,2.5) [2.5,3.5) [2.5,3.5) [2.5,3.5)
## [73] [2.5,3.5) [2.5,3.5) [2.5,3.5) [2.5,3.5) [2.5,3.5) [2.5,3.5)
## [79] [2.5,3.5) [2.5,3.5) [-Inf,2.5) [-Inf,2.5) [2.5,3.5) [2.5,3.5)
## [85] [2.5,3.5) [2.5,3.5) [2.5,3.5) [-Inf,2.5) [2.5,3.5) [2.5,3.5)
## [91] [2.5,3.5) [2.5,3.5) [2.5,3.5) [-Inf,2.5) [2.5,3.5) [2.5,3.5)
## [97] [2.5,3.5) [2.5,3.5) [2.5,3.5) [2.5,3.5) [2.5,3.5) [2.5,3.5)
## [103] [2.5,3.5) [2.5,3.5) [2.5,3.5) [2.5,3.5) [2.5,3.5) [2.5,3.5)
## [109] [2.5,3.5) [3.5, Inf] [2.5,3.5) [2.5,3.5) [2.5,3.5) [2.5,3.5)
## [115] [2.5,3.5) [2.5,3.5) [2.5,3.5) [3.5, Inf] [2.5,3.5) [-Inf,2.5)
## [121] [2.5,3.5) [2.5,3.5) [2.5,3.5) [2.5,3.5) [2.5,3.5) [2.5,3.5)
## [127] [2.5,3.5) [2.5,3.5) [2.5,3.5) [2.5,3.5) [2.5,3.5) [3.5, Inf]
## [133] [2.5,3.5) [2.5,3.5) [2.5,3.5) [2.5,3.5) [2.5,3.5) [2.5,3.5)
## [139] [2.5,3.5) [2.5,3.5) [2.5,3.5) [2.5,3.5) [2.5,3.5) [2.5,3.5)
## [145] [2.5,3.5) [2.5,3.5) [2.5,3.5) [2.5,3.5) [2.5,3.5) [2.5,3.5)
## attr(,"discretized:breaks")
## [1] -Inf 2.5 3.5 Inf
## attr(,"discretized:method")
## [1] fixed
## Levels: [-Inf,2.5) [2.5,3.5) [3.5, Inf]table(iris_interval)## iris_interval
## [2,2.8) [2.8,3.6) [3.6,4.4]
## 47 88 15Interval делит диапазон значений на N равных частей. Метод Interval очень страдает от выбросов.
Frequency делит данные так, чтобы в каждую корзину попало примерно одинаковое количество наблюдений.
Cluster использует алгоритм k-means для поиска естественных групп в данных.
Задание 4
Установите пакет Boruta и проведите выбор признаков для набора данных data(“Ozone”) [4, 5, 6]. Построить график boxplot, сделать выводы.
library(Boruta)
library(mlbench)
data("Ozone")
# V4 — целевая переменная (Daily maximum one-hour average ozone reading)
Ozone <- Ozone[complete.cases(Ozone$V4),]
boruta_result <- Boruta(V4 ~ ., data=Ozone)
print(boruta_result)## Boruta performed 99 iterations in 32.89965 secs.
## 6 attributes confirmed important: V10, V12, V5, V7, V8 and 1 more;
## 5 attributes confirmed unimportant: V11, V13, V2, V3, V6;
## 1 tentative attributes left: V1;plot(boruta_result, sort=FALSE)
В наборе Ozone наиболее важными признаками (зелеными) оказались признаки V5, V7-V10 и V12.
Признаки V2, V3, V6, V11, V13 были классифицированы как неважные, для признака V1 важность признака определить не удалось.
Вывод
В этой лабораторной работе мы продолжили изучение методов выбора признаков, используя различные пакеты R. Полученные результаты позволяют сравнить различные подходы к выявлению наиболее важных признаков для задач классификации и прогнозирования.