Лабораторная работа: отбор и дискретизация признаков
Сухарева Юлия, 233-331
2026-10-09
1. Графический разведочный анализ данных с использованием CARET
Установим и подключим пакет caret.
library(caret)## Загрузка требуемого пакета: ggplot2## Загрузка требуемого пакета: latticeВыведем список доступных моделей:
names(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"Создадим набор данных согласно заданию:
set.seed(123)
x <- matrix(rnorm(50 * 5), ncol = 5)
y <- factor(rep(c("A", "B"), 25))Boxplot
jpeg(
"feature_box.jpg",
width = 1200,
height = 900,
res = 150
)
featurePlot(
x = x,
y = y,
plot = "box"
)
dev.off()## png
## 2Для отображения графика в HTML:
featurePlot(
x = x,
y = y,
plot = "box"
)
Strip plot
jpeg(
"feature_strip.jpg",
width = 1200,
height = 900,
res = 150
)
featurePlot(
x = x,
y = y,
plot = "strip",
jitter = TRUE
)
dev.off()## png
## 2featurePlot(
x = x,
y = y,
plot = "strip",
jitter = TRUE
)
Density plot
jpeg(
"feature_density.jpg",
width = 1200,
height = 900,
res = 150
)
featurePlot(
x = x,
y = y,
plot = "density"
)
dev.off()## png
## 2featurePlot(
x = x,
y = y,
plot = "density"
)
Попарные зависимости признаков
jpeg(
"feature_pairs.jpg",
width = 1200,
height = 900,
res = 150
)
featurePlot(
x = x,
y = y,
plot = "pairs"
)
dev.off()## png
## 2featurePlot(
x = x,
y = y,
plot = "pairs"
)
Вывод
В результате графического разведочного анализа были построены графики
boxplot, strip plot, распределения плотности и попарных зависимостей
признаков с использованием функции featurePlot(). Поскольку
исходные признаки были случайно сгенерированы из нормального
распределения, выраженного разделения классов A и B по признакам не
наблюдается. Графический анализ позволяет оценить распределение
признаков и различия между классами.
2. Определение важности признаков с использованием FSelector
Подключим пакет FSelector и загрузим набор данных
iris.
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 setosaInformation Gain
ig <- information.gain(
Species ~ .,
data = iris
)
ig## attr_importance
## Sepal.Length 0.4521286
## Sepal.Width 0.2672750
## Petal.Length 0.9402853
## Petal.Width 0.9554360Gain Ratio
gr <- gain.ratio(
Species ~ .,
data = iris
)
gr## attr_importance
## Sepal.Length 0.4196464
## Sepal.Width 0.2472972
## Petal.Length 0.8584937
## Petal.Width 0.8713692Chi-Squared
chi <- chi.squared(
Species ~ .,
data = iris
)
chi## attr_importance
## Sepal.Length 0.6288067
## Sepal.Width 0.4922162
## Petal.Length 0.9346311
## Petal.Width 0.9432359Symmetrical Uncertainty
su <- symmetrical.uncertainty(
Species ~ .,
data = iris
)
su## attr_importance
## Sepal.Length 0.4155563
## Sepal.Width 0.2452743
## Petal.Length 0.8571872
## Petal.Width 0.8705214Вывод
В результате применения методов пакета FSelector была
определена информативность признаков набора iris
относительно целевой переменной Species. Полученные
значения показывают, что признаки имеют различную степень важности для
классификации. Наибольшую информативность имеют признаки
Petal.Length и Petal.Width, тогда как признаки
Sepal.Length и Sepal.Width имеют меньшую
информативность. Таким образом, размеры лепестка являются наиболее
существенными признаками для определения вида ириса.
3. Дискретизация непрерывной переменной с использованием arules
Подключим пакет arules и загрузим набор данных
iris.
library(arules)## Загрузка требуемого пакета: Matrix##
## Присоединяю пакет: 'arules'## Следующие объекты скрыты от 'package:base':
##
## abbreviate, writedata(iris)
x <- iris$Sepal.LengthМетод interval
d_interval <- discretize(
x,
method = "interval",
breaks = 3
)
d_interval## [1] [4.3,5.5) [4.3,5.5) [4.3,5.5) [4.3,5.5) [4.3,5.5) [4.3,5.5) [4.3,5.5)
## [8] [4.3,5.5) [4.3,5.5) [4.3,5.5) [4.3,5.5) [4.3,5.5) [4.3,5.5) [4.3,5.5)
## [15] [5.5,6.7) [5.5,6.7) [4.3,5.5) [4.3,5.5) [5.5,6.7) [4.3,5.5) [4.3,5.5)
## [22] [4.3,5.5) [4.3,5.5) [4.3,5.5) [4.3,5.5) [4.3,5.5) [4.3,5.5) [4.3,5.5)
## [29] [4.3,5.5) [4.3,5.5) [4.3,5.5) [4.3,5.5) [4.3,5.5) [5.5,6.7) [4.3,5.5)
## [36] [4.3,5.5) [5.5,6.7) [4.3,5.5) [4.3,5.5) [4.3,5.5) [4.3,5.5) [4.3,5.5)
## [43] [4.3,5.5) [4.3,5.5) [4.3,5.5) [4.3,5.5) [4.3,5.5) [4.3,5.5) [4.3,5.5)
## [50] [4.3,5.5) [6.7,7.9] [5.5,6.7) [6.7,7.9] [5.5,6.7) [5.5,6.7) [5.5,6.7)
## [57] [5.5,6.7) [4.3,5.5) [5.5,6.7) [4.3,5.5) [4.3,5.5) [5.5,6.7) [5.5,6.7)
## [64] [5.5,6.7) [5.5,6.7) [6.7,7.9] [5.5,6.7) [5.5,6.7) [5.5,6.7) [5.5,6.7)
## [71] [5.5,6.7) [5.5,6.7) [5.5,6.7) [5.5,6.7) [5.5,6.7) [5.5,6.7) [6.7,7.9]
## [78] [6.7,7.9] [5.5,6.7) [5.5,6.7) [5.5,6.7) [5.5,6.7) [5.5,6.7) [5.5,6.7)
## [85] [4.3,5.5) [5.5,6.7) [6.7,7.9] [5.5,6.7) [5.5,6.7) [5.5,6.7) [5.5,6.7)
## [92] [5.5,6.7) [5.5,6.7) [4.3,5.5) [5.5,6.7) [5.5,6.7) [5.5,6.7) [5.5,6.7)
## [99] [4.3,5.5) [5.5,6.7) [5.5,6.7) [5.5,6.7) [6.7,7.9] [5.5,6.7) [5.5,6.7)
## [106] [6.7,7.9] [4.3,5.5) [6.7,7.9] [6.7,7.9] [6.7,7.9] [5.5,6.7) [5.5,6.7)
## [113] [6.7,7.9] [5.5,6.7) [5.5,6.7) [5.5,6.7) [5.5,6.7) [6.7,7.9] [6.7,7.9]
## [120] [5.5,6.7) [6.7,7.9] [5.5,6.7) [6.7,7.9] [5.5,6.7) [6.7,7.9] [6.7,7.9]
## [127] [5.5,6.7) [5.5,6.7) [5.5,6.7) [6.7,7.9] [6.7,7.9] [6.7,7.9] [5.5,6.7)
## [134] [5.5,6.7) [5.5,6.7) [6.7,7.9] [5.5,6.7) [5.5,6.7) [5.5,6.7) [6.7,7.9]
## [141] [6.7,7.9] [6.7,7.9] [5.5,6.7) [6.7,7.9] [6.7,7.9] [6.7,7.9] [5.5,6.7)
## [148] [5.5,6.7) [5.5,6.7) [5.5,6.7)
## attr(,"discretized:breaks")
## [1] 4.3 5.5 6.7 7.9
## attr(,"discretized:method")
## [1] interval
## Levels: [4.3,5.5) [5.5,6.7) [6.7,7.9]Метод frequency
d_frequency <- discretize(
x,
method = "frequency",
breaks = 3
)
d_frequency## [1] [4.3,5.4) [4.3,5.4) [4.3,5.4) [4.3,5.4) [4.3,5.4) [5.4,6.3) [4.3,5.4)
## [8] [4.3,5.4) [4.3,5.4) [4.3,5.4) [5.4,6.3) [4.3,5.4) [4.3,5.4) [4.3,5.4)
## [15] [5.4,6.3) [5.4,6.3) [5.4,6.3) [4.3,5.4) [5.4,6.3) [4.3,5.4) [5.4,6.3)
## [22] [4.3,5.4) [4.3,5.4) [4.3,5.4) [4.3,5.4) [4.3,5.4) [4.3,5.4) [4.3,5.4)
## [29] [4.3,5.4) [4.3,5.4) [4.3,5.4) [5.4,6.3) [4.3,5.4) [5.4,6.3) [4.3,5.4)
## [36] [4.3,5.4) [5.4,6.3) [4.3,5.4) [4.3,5.4) [4.3,5.4) [4.3,5.4) [4.3,5.4)
## [43] [4.3,5.4) [4.3,5.4) [4.3,5.4) [4.3,5.4) [4.3,5.4) [4.3,5.4) [4.3,5.4)
## [50] [4.3,5.4) [6.3,7.9] [6.3,7.9] [6.3,7.9] [5.4,6.3) [6.3,7.9] [5.4,6.3)
## [57] [6.3,7.9] [4.3,5.4) [6.3,7.9] [4.3,5.4) [4.3,5.4) [5.4,6.3) [5.4,6.3)
## [64] [5.4,6.3) [5.4,6.3) [6.3,7.9] [5.4,6.3) [5.4,6.3) [5.4,6.3) [5.4,6.3)
## [71] [5.4,6.3) [5.4,6.3) [6.3,7.9] [5.4,6.3) [6.3,7.9] [6.3,7.9] [6.3,7.9]
## [78] [6.3,7.9] [5.4,6.3) [5.4,6.3) [5.4,6.3) [5.4,6.3) [5.4,6.3) [5.4,6.3)
## [85] [5.4,6.3) [5.4,6.3) [6.3,7.9] [6.3,7.9] [5.4,6.3) [5.4,6.3) [5.4,6.3)
## [92] [5.4,6.3) [5.4,6.3) [4.3,5.4) [5.4,6.3) [5.4,6.3) [5.4,6.3) [5.4,6.3)
## [99] [4.3,5.4) [5.4,6.3) [6.3,7.9] [5.4,6.3) [6.3,7.9] [6.3,7.9] [6.3,7.9]
## [106] [6.3,7.9] [4.3,5.4) [6.3,7.9] [6.3,7.9] [6.3,7.9] [6.3,7.9] [6.3,7.9]
## [113] [6.3,7.9] [5.4,6.3) [5.4,6.3) [6.3,7.9] [6.3,7.9] [6.3,7.9] [6.3,7.9]
## [120] [5.4,6.3) [6.3,7.9] [5.4,6.3) [6.3,7.9] [6.3,7.9] [6.3,7.9] [6.3,7.9]
## [127] [5.4,6.3) [5.4,6.3) [6.3,7.9] [6.3,7.9] [6.3,7.9] [6.3,7.9] [6.3,7.9]
## [134] [6.3,7.9] [5.4,6.3) [6.3,7.9] [6.3,7.9] [6.3,7.9] [5.4,6.3) [6.3,7.9]
## [141] [6.3,7.9] [6.3,7.9] [5.4,6.3) [6.3,7.9] [6.3,7.9] [6.3,7.9] [6.3,7.9]
## [148] [6.3,7.9] [5.4,6.3) [5.4,6.3)
## attr(,"discretized:breaks")
## [1] 4.3 5.4 6.3 7.9
## attr(,"discretized:method")
## [1] frequency
## Levels: [4.3,5.4) [5.4,6.3) [6.3,7.9]Метод cluster
set.seed(123)
d_cluster <- discretize(
x,
method = "cluster",
breaks = 3
)
d_cluster## [1] [4.3,5.37) [4.3,5.37) [4.3,5.37) [4.3,5.37) [4.3,5.37) [5.37,6.36)
## [7] [4.3,5.37) [4.3,5.37) [4.3,5.37) [4.3,5.37) [5.37,6.36) [4.3,5.37)
## [13] [4.3,5.37) [4.3,5.37) [5.37,6.36) [5.37,6.36) [5.37,6.36) [4.3,5.37)
## [19] [5.37,6.36) [4.3,5.37) [5.37,6.36) [4.3,5.37) [4.3,5.37) [4.3,5.37)
## [25] [4.3,5.37) [4.3,5.37) [4.3,5.37) [4.3,5.37) [4.3,5.37) [4.3,5.37)
## [31] [4.3,5.37) [5.37,6.36) [4.3,5.37) [5.37,6.36) [4.3,5.37) [4.3,5.37)
## [37] [5.37,6.36) [4.3,5.37) [4.3,5.37) [4.3,5.37) [4.3,5.37) [4.3,5.37)
## [43] [4.3,5.37) [4.3,5.37) [4.3,5.37) [4.3,5.37) [4.3,5.37) [4.3,5.37)
## [49] [4.3,5.37) [4.3,5.37) [6.36,7.9] [6.36,7.9] [6.36,7.9] [5.37,6.36)
## [55] [6.36,7.9] [5.37,6.36) [5.37,6.36) [4.3,5.37) [6.36,7.9] [4.3,5.37)
## [61] [4.3,5.37) [5.37,6.36) [5.37,6.36) [5.37,6.36) [5.37,6.36) [6.36,7.9]
## [67] [5.37,6.36) [5.37,6.36) [5.37,6.36) [5.37,6.36) [5.37,6.36) [5.37,6.36)
## [73] [5.37,6.36) [5.37,6.36) [6.36,7.9] [6.36,7.9] [6.36,7.9] [6.36,7.9]
## [79] [5.37,6.36) [5.37,6.36) [5.37,6.36) [5.37,6.36) [5.37,6.36) [5.37,6.36)
## [85] [5.37,6.36) [5.37,6.36) [6.36,7.9] [5.37,6.36) [5.37,6.36) [5.37,6.36)
## [91] [5.37,6.36) [5.37,6.36) [5.37,6.36) [4.3,5.37) [5.37,6.36) [5.37,6.36)
## [97] [5.37,6.36) [5.37,6.36) [4.3,5.37) [5.37,6.36) [5.37,6.36) [5.37,6.36)
## [103] [6.36,7.9] [5.37,6.36) [6.36,7.9] [6.36,7.9] [4.3,5.37) [6.36,7.9]
## [109] [6.36,7.9] [6.36,7.9] [6.36,7.9] [6.36,7.9] [6.36,7.9] [5.37,6.36)
## [115] [5.37,6.36) [6.36,7.9] [6.36,7.9] [6.36,7.9] [6.36,7.9] [5.37,6.36)
## [121] [6.36,7.9] [5.37,6.36) [6.36,7.9] [5.37,6.36) [6.36,7.9] [6.36,7.9]
## [127] [5.37,6.36) [5.37,6.36) [6.36,7.9] [6.36,7.9] [6.36,7.9] [6.36,7.9]
## [133] [6.36,7.9] [5.37,6.36) [5.37,6.36) [6.36,7.9] [5.37,6.36) [6.36,7.9]
## [139] [5.37,6.36) [6.36,7.9] [6.36,7.9] [6.36,7.9] [5.37,6.36) [6.36,7.9]
## [145] [6.36,7.9] [6.36,7.9] [5.37,6.36) [6.36,7.9] [5.37,6.36) [5.37,6.36)
## attr(,"discretized:breaks")
## [1] 4.300000 5.371213 6.361482 7.900000
## attr(,"discretized:method")
## [1] cluster
## Levels: [4.3,5.37) [5.37,6.36) [6.36,7.9]Метод fixed
d_fixed <- discretize(
x,
method = "fixed",
breaks = c(-Inf, 5, 6, Inf)
)
d_fixed## [1] [5,6) [-Inf,5) [-Inf,5) [-Inf,5) [5,6) [5,6) [-Inf,5) [5,6)
## [9] [-Inf,5) [-Inf,5) [5,6) [-Inf,5) [-Inf,5) [-Inf,5) [5,6) [5,6)
## [17] [5,6) [5,6) [5,6) [5,6) [5,6) [5,6) [-Inf,5) [5,6)
## [25] [-Inf,5) [5,6) [5,6) [5,6) [5,6) [-Inf,5) [-Inf,5) [5,6)
## [33] [5,6) [5,6) [-Inf,5) [5,6) [5,6) [-Inf,5) [-Inf,5) [5,6)
## [41] [5,6) [-Inf,5) [-Inf,5) [5,6) [5,6) [-Inf,5) [5,6) [-Inf,5)
## [49] [5,6) [5,6) [6, Inf] [6, Inf] [6, Inf] [5,6) [6, Inf] [5,6)
## [57] [6, Inf] [-Inf,5) [6, Inf] [5,6) [5,6) [5,6) [6, Inf] [6, Inf]
## [65] [5,6) [6, Inf] [5,6) [5,6) [6, Inf] [5,6) [5,6) [6, Inf]
## [73] [6, Inf] [6, Inf] [6, Inf] [6, Inf] [6, Inf] [6, Inf] [6, Inf] [5,6)
## [81] [5,6) [5,6) [5,6) [6, Inf] [5,6) [6, Inf] [6, Inf] [6, Inf]
## [89] [5,6) [5,6) [5,6) [6, Inf] [5,6) [5,6) [5,6) [5,6)
## [97] [5,6) [6, Inf] [5,6) [5,6) [6, Inf] [5,6) [6, Inf] [6, Inf]
## [105] [6, Inf] [6, Inf] [-Inf,5) [6, Inf] [6, Inf] [6, Inf] [6, Inf] [6, Inf]
## [113] [6, Inf] [5,6) [5,6) [6, Inf] [6, Inf] [6, Inf] [6, Inf] [6, Inf]
## [121] [6, Inf] [5,6) [6, Inf] [6, Inf] [6, Inf] [6, Inf] [6, Inf] [6, Inf]
## [129] [6, Inf] [6, Inf] [6, Inf] [6, Inf] [6, Inf] [6, Inf] [6, Inf] [6, Inf]
## [137] [6, Inf] [6, Inf] [6, Inf] [6, Inf] [6, Inf] [6, Inf] [5,6) [6, Inf]
## [145] [6, Inf] [6, Inf] [6, Inf] [6, Inf] [6, Inf] [5,6)
## attr(,"discretized:breaks")
## [1] -Inf 5 6 Inf
## attr(,"discretized:method")
## [1] fixed
## Levels: [-Inf,5) [5,6) [6, Inf]Сравнение результатов
data.frame(
Original = x,
Interval = d_interval,
Frequency = d_frequency,
Cluster = d_cluster,
Fixed = d_fixed
)## Original Interval Frequency Cluster Fixed
## 1 5.1 [4.3,5.5) [4.3,5.4) [4.3,5.37) [5,6)
## 2 4.9 [4.3,5.5) [4.3,5.4) [4.3,5.37) [-Inf,5)
## 3 4.7 [4.3,5.5) [4.3,5.4) [4.3,5.37) [-Inf,5)
## 4 4.6 [4.3,5.5) [4.3,5.4) [4.3,5.37) [-Inf,5)
## 5 5.0 [4.3,5.5) [4.3,5.4) [4.3,5.37) [5,6)
## 6 5.4 [4.3,5.5) [5.4,6.3) [5.37,6.36) [5,6)
## 7 4.6 [4.3,5.5) [4.3,5.4) [4.3,5.37) [-Inf,5)
## 8 5.0 [4.3,5.5) [4.3,5.4) [4.3,5.37) [5,6)
## 9 4.4 [4.3,5.5) [4.3,5.4) [4.3,5.37) [-Inf,5)
## 10 4.9 [4.3,5.5) [4.3,5.4) [4.3,5.37) [-Inf,5)
## 11 5.4 [4.3,5.5) [5.4,6.3) [5.37,6.36) [5,6)
## 12 4.8 [4.3,5.5) [4.3,5.4) [4.3,5.37) [-Inf,5)
## 13 4.8 [4.3,5.5) [4.3,5.4) [4.3,5.37) [-Inf,5)
## 14 4.3 [4.3,5.5) [4.3,5.4) [4.3,5.37) [-Inf,5)
## 15 5.8 [5.5,6.7) [5.4,6.3) [5.37,6.36) [5,6)
## 16 5.7 [5.5,6.7) [5.4,6.3) [5.37,6.36) [5,6)
## 17 5.4 [4.3,5.5) [5.4,6.3) [5.37,6.36) [5,6)
## 18 5.1 [4.3,5.5) [4.3,5.4) [4.3,5.37) [5,6)
## 19 5.7 [5.5,6.7) [5.4,6.3) [5.37,6.36) [5,6)
## 20 5.1 [4.3,5.5) [4.3,5.4) [4.3,5.37) [5,6)
## 21 5.4 [4.3,5.5) [5.4,6.3) [5.37,6.36) [5,6)
## 22 5.1 [4.3,5.5) [4.3,5.4) [4.3,5.37) [5,6)
## 23 4.6 [4.3,5.5) [4.3,5.4) [4.3,5.37) [-Inf,5)
## 24 5.1 [4.3,5.5) [4.3,5.4) [4.3,5.37) [5,6)
## 25 4.8 [4.3,5.5) [4.3,5.4) [4.3,5.37) [-Inf,5)
## 26 5.0 [4.3,5.5) [4.3,5.4) [4.3,5.37) [5,6)
## 27 5.0 [4.3,5.5) [4.3,5.4) [4.3,5.37) [5,6)
## 28 5.2 [4.3,5.5) [4.3,5.4) [4.3,5.37) [5,6)
## 29 5.2 [4.3,5.5) [4.3,5.4) [4.3,5.37) [5,6)
## 30 4.7 [4.3,5.5) [4.3,5.4) [4.3,5.37) [-Inf,5)
## 31 4.8 [4.3,5.5) [4.3,5.4) [4.3,5.37) [-Inf,5)
## 32 5.4 [4.3,5.5) [5.4,6.3) [5.37,6.36) [5,6)
## 33 5.2 [4.3,5.5) [4.3,5.4) [4.3,5.37) [5,6)
## 34 5.5 [5.5,6.7) [5.4,6.3) [5.37,6.36) [5,6)
## 35 4.9 [4.3,5.5) [4.3,5.4) [4.3,5.37) [-Inf,5)
## 36 5.0 [4.3,5.5) [4.3,5.4) [4.3,5.37) [5,6)
## 37 5.5 [5.5,6.7) [5.4,6.3) [5.37,6.36) [5,6)
## 38 4.9 [4.3,5.5) [4.3,5.4) [4.3,5.37) [-Inf,5)
## 39 4.4 [4.3,5.5) [4.3,5.4) [4.3,5.37) [-Inf,5)
## 40 5.1 [4.3,5.5) [4.3,5.4) [4.3,5.37) [5,6)
## 41 5.0 [4.3,5.5) [4.3,5.4) [4.3,5.37) [5,6)
## 42 4.5 [4.3,5.5) [4.3,5.4) [4.3,5.37) [-Inf,5)
## 43 4.4 [4.3,5.5) [4.3,5.4) [4.3,5.37) [-Inf,5)
## 44 5.0 [4.3,5.5) [4.3,5.4) [4.3,5.37) [5,6)
## 45 5.1 [4.3,5.5) [4.3,5.4) [4.3,5.37) [5,6)
## 46 4.8 [4.3,5.5) [4.3,5.4) [4.3,5.37) [-Inf,5)
## 47 5.1 [4.3,5.5) [4.3,5.4) [4.3,5.37) [5,6)
## 48 4.6 [4.3,5.5) [4.3,5.4) [4.3,5.37) [-Inf,5)
## 49 5.3 [4.3,5.5) [4.3,5.4) [4.3,5.37) [5,6)
## 50 5.0 [4.3,5.5) [4.3,5.4) [4.3,5.37) [5,6)
## 51 7.0 [6.7,7.9] [6.3,7.9] [6.36,7.9] [6, Inf]
## 52 6.4 [5.5,6.7) [6.3,7.9] [6.36,7.9] [6, Inf]
## 53 6.9 [6.7,7.9] [6.3,7.9] [6.36,7.9] [6, Inf]
## 54 5.5 [5.5,6.7) [5.4,6.3) [5.37,6.36) [5,6)
## 55 6.5 [5.5,6.7) [6.3,7.9] [6.36,7.9] [6, Inf]
## 56 5.7 [5.5,6.7) [5.4,6.3) [5.37,6.36) [5,6)
## 57 6.3 [5.5,6.7) [6.3,7.9] [5.37,6.36) [6, Inf]
## 58 4.9 [4.3,5.5) [4.3,5.4) [4.3,5.37) [-Inf,5)
## 59 6.6 [5.5,6.7) [6.3,7.9] [6.36,7.9] [6, Inf]
## 60 5.2 [4.3,5.5) [4.3,5.4) [4.3,5.37) [5,6)
## 61 5.0 [4.3,5.5) [4.3,5.4) [4.3,5.37) [5,6)
## 62 5.9 [5.5,6.7) [5.4,6.3) [5.37,6.36) [5,6)
## 63 6.0 [5.5,6.7) [5.4,6.3) [5.37,6.36) [6, Inf]
## 64 6.1 [5.5,6.7) [5.4,6.3) [5.37,6.36) [6, Inf]
## 65 5.6 [5.5,6.7) [5.4,6.3) [5.37,6.36) [5,6)
## 66 6.7 [6.7,7.9] [6.3,7.9] [6.36,7.9] [6, Inf]
## 67 5.6 [5.5,6.7) [5.4,6.3) [5.37,6.36) [5,6)
## 68 5.8 [5.5,6.7) [5.4,6.3) [5.37,6.36) [5,6)
## 69 6.2 [5.5,6.7) [5.4,6.3) [5.37,6.36) [6, Inf]
## 70 5.6 [5.5,6.7) [5.4,6.3) [5.37,6.36) [5,6)
## 71 5.9 [5.5,6.7) [5.4,6.3) [5.37,6.36) [5,6)
## 72 6.1 [5.5,6.7) [5.4,6.3) [5.37,6.36) [6, Inf]
## 73 6.3 [5.5,6.7) [6.3,7.9] [5.37,6.36) [6, Inf]
## 74 6.1 [5.5,6.7) [5.4,6.3) [5.37,6.36) [6, Inf]
## 75 6.4 [5.5,6.7) [6.3,7.9] [6.36,7.9] [6, Inf]
## 76 6.6 [5.5,6.7) [6.3,7.9] [6.36,7.9] [6, Inf]
## 77 6.8 [6.7,7.9] [6.3,7.9] [6.36,7.9] [6, Inf]
## 78 6.7 [6.7,7.9] [6.3,7.9] [6.36,7.9] [6, Inf]
## 79 6.0 [5.5,6.7) [5.4,6.3) [5.37,6.36) [6, Inf]
## 80 5.7 [5.5,6.7) [5.4,6.3) [5.37,6.36) [5,6)
## 81 5.5 [5.5,6.7) [5.4,6.3) [5.37,6.36) [5,6)
## 82 5.5 [5.5,6.7) [5.4,6.3) [5.37,6.36) [5,6)
## 83 5.8 [5.5,6.7) [5.4,6.3) [5.37,6.36) [5,6)
## 84 6.0 [5.5,6.7) [5.4,6.3) [5.37,6.36) [6, Inf]
## 85 5.4 [4.3,5.5) [5.4,6.3) [5.37,6.36) [5,6)
## 86 6.0 [5.5,6.7) [5.4,6.3) [5.37,6.36) [6, Inf]
## 87 6.7 [6.7,7.9] [6.3,7.9] [6.36,7.9] [6, Inf]
## 88 6.3 [5.5,6.7) [6.3,7.9] [5.37,6.36) [6, Inf]
## 89 5.6 [5.5,6.7) [5.4,6.3) [5.37,6.36) [5,6)
## 90 5.5 [5.5,6.7) [5.4,6.3) [5.37,6.36) [5,6)
## 91 5.5 [5.5,6.7) [5.4,6.3) [5.37,6.36) [5,6)
## 92 6.1 [5.5,6.7) [5.4,6.3) [5.37,6.36) [6, Inf]
## 93 5.8 [5.5,6.7) [5.4,6.3) [5.37,6.36) [5,6)
## 94 5.0 [4.3,5.5) [4.3,5.4) [4.3,5.37) [5,6)
## 95 5.6 [5.5,6.7) [5.4,6.3) [5.37,6.36) [5,6)
## 96 5.7 [5.5,6.7) [5.4,6.3) [5.37,6.36) [5,6)
## 97 5.7 [5.5,6.7) [5.4,6.3) [5.37,6.36) [5,6)
## 98 6.2 [5.5,6.7) [5.4,6.3) [5.37,6.36) [6, Inf]
## 99 5.1 [4.3,5.5) [4.3,5.4) [4.3,5.37) [5,6)
## 100 5.7 [5.5,6.7) [5.4,6.3) [5.37,6.36) [5,6)
## 101 6.3 [5.5,6.7) [6.3,7.9] [5.37,6.36) [6, Inf]
## 102 5.8 [5.5,6.7) [5.4,6.3) [5.37,6.36) [5,6)
## 103 7.1 [6.7,7.9] [6.3,7.9] [6.36,7.9] [6, Inf]
## 104 6.3 [5.5,6.7) [6.3,7.9] [5.37,6.36) [6, Inf]
## 105 6.5 [5.5,6.7) [6.3,7.9] [6.36,7.9] [6, Inf]
## 106 7.6 [6.7,7.9] [6.3,7.9] [6.36,7.9] [6, Inf]
## 107 4.9 [4.3,5.5) [4.3,5.4) [4.3,5.37) [-Inf,5)
## 108 7.3 [6.7,7.9] [6.3,7.9] [6.36,7.9] [6, Inf]
## 109 6.7 [6.7,7.9] [6.3,7.9] [6.36,7.9] [6, Inf]
## 110 7.2 [6.7,7.9] [6.3,7.9] [6.36,7.9] [6, Inf]
## 111 6.5 [5.5,6.7) [6.3,7.9] [6.36,7.9] [6, Inf]
## 112 6.4 [5.5,6.7) [6.3,7.9] [6.36,7.9] [6, Inf]
## 113 6.8 [6.7,7.9] [6.3,7.9] [6.36,7.9] [6, Inf]
## 114 5.7 [5.5,6.7) [5.4,6.3) [5.37,6.36) [5,6)
## 115 5.8 [5.5,6.7) [5.4,6.3) [5.37,6.36) [5,6)
## 116 6.4 [5.5,6.7) [6.3,7.9] [6.36,7.9] [6, Inf]
## 117 6.5 [5.5,6.7) [6.3,7.9] [6.36,7.9] [6, Inf]
## 118 7.7 [6.7,7.9] [6.3,7.9] [6.36,7.9] [6, Inf]
## 119 7.7 [6.7,7.9] [6.3,7.9] [6.36,7.9] [6, Inf]
## 120 6.0 [5.5,6.7) [5.4,6.3) [5.37,6.36) [6, Inf]
## 121 6.9 [6.7,7.9] [6.3,7.9] [6.36,7.9] [6, Inf]
## 122 5.6 [5.5,6.7) [5.4,6.3) [5.37,6.36) [5,6)
## 123 7.7 [6.7,7.9] [6.3,7.9] [6.36,7.9] [6, Inf]
## 124 6.3 [5.5,6.7) [6.3,7.9] [5.37,6.36) [6, Inf]
## 125 6.7 [6.7,7.9] [6.3,7.9] [6.36,7.9] [6, Inf]
## 126 7.2 [6.7,7.9] [6.3,7.9] [6.36,7.9] [6, Inf]
## 127 6.2 [5.5,6.7) [5.4,6.3) [5.37,6.36) [6, Inf]
## 128 6.1 [5.5,6.7) [5.4,6.3) [5.37,6.36) [6, Inf]
## 129 6.4 [5.5,6.7) [6.3,7.9] [6.36,7.9] [6, Inf]
## 130 7.2 [6.7,7.9] [6.3,7.9] [6.36,7.9] [6, Inf]
## 131 7.4 [6.7,7.9] [6.3,7.9] [6.36,7.9] [6, Inf]
## 132 7.9 [6.7,7.9] [6.3,7.9] [6.36,7.9] [6, Inf]
## 133 6.4 [5.5,6.7) [6.3,7.9] [6.36,7.9] [6, Inf]
## 134 6.3 [5.5,6.7) [6.3,7.9] [5.37,6.36) [6, Inf]
## 135 6.1 [5.5,6.7) [5.4,6.3) [5.37,6.36) [6, Inf]
## 136 7.7 [6.7,7.9] [6.3,7.9] [6.36,7.9] [6, Inf]
## 137 6.3 [5.5,6.7) [6.3,7.9] [5.37,6.36) [6, Inf]
## 138 6.4 [5.5,6.7) [6.3,7.9] [6.36,7.9] [6, Inf]
## 139 6.0 [5.5,6.7) [5.4,6.3) [5.37,6.36) [6, Inf]
## 140 6.9 [6.7,7.9] [6.3,7.9] [6.36,7.9] [6, Inf]
## 141 6.7 [6.7,7.9] [6.3,7.9] [6.36,7.9] [6, Inf]
## 142 6.9 [6.7,7.9] [6.3,7.9] [6.36,7.9] [6, Inf]
## 143 5.8 [5.5,6.7) [5.4,6.3) [5.37,6.36) [5,6)
## 144 6.8 [6.7,7.9] [6.3,7.9] [6.36,7.9] [6, Inf]
## 145 6.7 [6.7,7.9] [6.3,7.9] [6.36,7.9] [6, Inf]
## 146 6.7 [6.7,7.9] [6.3,7.9] [6.36,7.9] [6, Inf]
## 147 6.3 [5.5,6.7) [6.3,7.9] [5.37,6.36) [6, Inf]
## 148 6.5 [5.5,6.7) [6.3,7.9] [6.36,7.9] [6, Inf]
## 149 6.2 [5.5,6.7) [5.4,6.3) [5.37,6.36) [6, Inf]
## 150 5.9 [5.5,6.7) [5.4,6.3) [5.37,6.36) [5,6)Вывод
В результате выполнения дискретизации непрерывная переменная
Sepal.Length была преобразована в категориальную четырьмя
различными способами. Метод interval формирует интервалы
одинаковой ширины, frequency — интервалы с примерно
одинаковым количеством наблюдений, cluster определяет
категории на основе кластеризации методом k-means, а fixed
использует заданные пользователем границы. В результате разные методы
приводят к различному разбиению исходной непрерывной переменной на
категории.
4. Выбор признаков с использованием Boruta
Подключим пакет Boruta и пакет mlbench,
содержащий набор данных Ozone.
library(Boruta)
library(mlbench)
data(Ozone)
ozo <- na.omit(Ozone)
head(ozo)## V1 V2 V3 V4 V5 V6 V7 V8 V9 V10 V11 V12 V13
## 5 1 5 1 5 5760 3 51 54 45.32 1450 25 57.02 60
## 6 1 6 2 6 5720 4 69 35 49.64 1568 15 53.78 60
## 7 1 7 3 4 5790 6 19 45 46.40 2631 -33 54.14 100
## 8 1 8 4 4 5790 3 25 55 52.70 554 -28 64.76 250
## 9 1 9 5 6 5700 3 73 41 48.02 2083 23 52.52 120
## 12 1 12 1 6 5720 3 44 51 54.32 111 9 63.14 150Проверим наличие пропущенных значений:
sum(is.na(ozo))## [1] 0Выполним выбор признаков:
set.seed(123)
boruta_result <- Boruta(
V4 ~ .,
data = ozo,
doTrace = 2
)## 1. run of importance source...## 2. run of importance source...## 3. run of importance source...## 4. run of importance source...## 5. run of importance source...## 6. run of importance source...## 7. run of importance source...## 8. run of importance source...## 9. run of importance source...## 10. run of importance source...## 11. run of importance source...## After 11 iterations, +0.17 secs:## confirmed 9 attributes: V1, V10, V11, V12, V13 and 4 more;## rejected 1 attribute: V3;## still have 2 attributes left.## 12. run of importance source...## 13. run of importance source...## 14. run of importance source...## 15. run of importance source...## After 15 iterations, +0.23 secs:## rejected 1 attribute: V6;## still have 1 attribute left.## 16. run of importance source...## 17. run of importance source...## 18. run of importance source...## After 18 iterations, +0.27 secs:## rejected 1 attribute: V2;## no more attributes left.print(boruta_result)## Boruta performed 18 iterations in 0.2735691 secs.
## 9 attributes confirmed important: V1, V10, V11, V12, V13 and 4 more;
## 3 attributes confirmed unimportant: V2, V3, V6;Итоговое решение
boruta_result$finalDecision## V1 V2 V3 V5 V6 V7 V8 V9
## Confirmed Rejected Rejected Confirmed Rejected Confirmed Confirmed Confirmed
## V10 V11 V12 V13
## Confirmed Confirmed Confirmed Confirmed
## Levels: Tentative Confirmed RejectedОтобранные признаки
getSelectedAttributes(boruta_result)## [1] "V1" "V5" "V7" "V8" "V9" "V10" "V11" "V12" "V13"Boxplot важности признаков
plot(
boruta_result,
las = 2
)
Вывод
В результате применения алгоритма Boruta была выполнена оценка
значимости признаков относительно целевой переменной V4.
Алгоритм сравнивает важность исходных признаков с важностью случайных
shadow-признаков. В результате признаки получают решения
Confirmed, Tentative или
Rejected. Подтверждённые признаки являются информативными
для рассматриваемой задачи, а отклонённые признаки не показали
достаточной значимости. На графике boxplot представлены распределения
важности признаков, полученные в процессе работы алгоритма.
5. Общий вывод
В ходе лабораторной работы были рассмотрены методы графического анализа, оценки важности и отбора признаков, а также дискретизации непрерывных данных.
С помощью пакета caret выполнен графический разведочный
анализ признаков с использованием функции featurePlot(). С
помощью пакета FSelector определена важность признаков для
классификации набора данных iris.
С использованием функции discretize() пакета
arules непрерывная переменная была преобразована в
категориальную четырьмя методами: interval,
frequency, cluster и fixed.
С помощью алгоритма Boruta выполнен выбор информативных признаков для
набора данных Ozone и построен boxplot важности
признаков.