caret::trainControl <-
function (method = "boot", number = ifelse(grepl("cv", method), 10, 25), repeats = ifelse(grepl("[d_]cv$", method), 1, NA), p = 0.75, search = "grid", initialWindow = NULL, horizon = 1, fixedWindow = TRUE, skip = 0, verboseIter = FALSE， returnData = TRUE, returnResamp = "final", savePredictions = FALSE,
classProbs = FALSE, summaryFunction = defaultSummary, selectionFunction = "best",
preProcOptions = list(thresh = 0.95, ICAcomp = 3, k = 5,
freqCut = 95/5, uniqueCut = 10, cutoff = 0.9), sampling = NULL,
index = NULL, indexOut = NULL, indexFinal = NULL, timingSamps = 0,
predictionBounds = rep(FALSE, 2), seeds = NA, adaptive = list(min = 5,
alpha = 0.05, method = "gls", complete = TRUE),
trim = FALSE, allowParallel = TRUE)
{
if (is.null(selectionFunction))
stop("null selectionFunction values not allowed")
if (!(returnResamp %in% c("all", "final", "none")))
stop("incorrect value of returnResamp")
if (length(predictionBounds) > 0 && length(predictionBounds) !=
2)
stop("'predictionBounds' should be a logical or numeric vector of length 2")
if (any(names(preProcOptions) == "method"))
stop("'method' cannot be specified here")
if (any(names(preProcOptions) == "x"))
stop("'x' cannot be specified here")
if (!is.na(repeats) & !(method %in% c("repeatedcv",
"adaptive_cv")))
warning("`repeats` has no meaning for this resampling method.",
call. = FALSE)
if (!(adaptive$method %in% c("gls", "BT")))
stop("incorrect value of adaptive$method")
if (adaptive$alpha < 1e-07 | adaptive$alpha > 1)
stop("incorrect value of adaptive$alpha")
if (grepl("adapt", method)) {
num <- if (method == "adaptive_cv")
number * repeats
else number
if (adaptive$min >= num)
stop(paste("adaptive$min should be less than",
num))
if (adaptive$min <= 1)
stop("adaptive$min should be greater than 1")
}
if (!(search %in% c("grid", "random")))
stop("`search` should be either 'grid' or 'random'")
if (method == "oob" & any(names(match.call()) == "summaryFunction")) {
warning("Custom summary measures cannot be computed for out-of-bag resampling. ",
"This value of `summaryFunction` will be ignored.",
call. = FALSE)
}
list(method = method, number = number, repeats = repeats,
search = search, p = p, initialWindow = initialWindow,
horizon = horizon, fixedWindow = fixedWindow, skip = skip,
verboseIter = verboseIter, returnData = returnData, returnResamp = returnResamp,
savePredictions = savePredictions, classProbs = classProbs,
summaryFunction = summaryFunction, selectionFunction = selectionFunction,
preProcOptions = preProcOptions, sampling = sampling,
index = index, indexOut = indexOut, indexFinal = indexFinal,
timingSamps = timingSamps, predictionBounds = predictionBounds,
seeds = seeds, adaptive = adaptive, trim = trim, allowParallel = allowParallel)
}
| trainControl | 所有参数详解 |
|---|---|
| method | 重抽样方法:Bootstrap(有放回随机抽样) 、Bootstrap632(有放回随机抽样扩展)、LOOCV(留一交叉验证)、LGOCV(蒙特卡罗交叉验证)、cv(k折交叉验证)、repeatedcv(重复的k折交叉验证)、optimism_boot(Efron, B., & Tibshirani, R. J. (1994). “An introduction to the bootstrap”, pages 249-252. CRC press.)、none(仅使用一个训练集拟合模型)、oob(袋外估计：随机森林、多元自适应回归样条、树模型、灵活判别分析、条件树) |
| number | 控制K折交叉验证的数目或者Bootstrap和LGOCV的抽样迭代次数 |
| repeats | 控制重复交叉验证的次数 |
| p | LGOCV：控制训练比例 |
| verboseIter | 输出训练日志的逻辑变量 |
| returnData | 逻辑变量，把数据保存到trainingData中（str(trainControl)查看） |
| search | search = grid(网格搜索)，random(随机搜索) |
| returnResamp | 包含以下值的字符串：final、all、none，设定有多少抽样性能度量被保存。 |
| classProbs | 是否计算类别概率 |
| summaryFunction | 根据重抽样计算模型性能的函数 |
| selectionFunction | 选择最优参数的函数 |
| index | 指定重抽样样本(使用相同的重抽样样本评估不同的算法、模型) |
| allowParallel | 是否允许并行 |
library(mlbench) #使用包中的数据 Warning message: 程辑包‘mlbench'是用R版本4.1.3 来建造的 > data(Sonar) > str(Sonar[, 1:10]) 'data.frame': 208 obs. of 10 variables: $ V1 : num 0.02 0.0453 0.0262 0.01 0.0762 0.0286 0.0317 0.0519 0.0223 0.0164 ... $ V2 : num 0.0371 0.0523 0.0582 0.0171 0.0666 0.0453 0.0956 0.0548 0.0375 0.0173 ... $ V3 : num 0.0428 0.0843 0.1099 0.0623 0.0481 ... $ V4 : num 0.0207 0.0689 0.1083 0.0205 0.0394 ... $ V5 : num 0.0954 0.1183 0.0974 0.0205 0.059 ... $ V6 : num 0.0986 0.2583 0.228 0.0368 0.0649 ... $ V7 : num 0.154 0.216 0.243 0.11 0.121 ... $ V8 : num 0.16 0.348 0.377 0.128 0.247 ... $ V9 : num 0.3109 0.3337 0.5598 0.0598 0.3564 ... $ V10: num 0.211 0.287 0.619 0.126 0.446 ...
数据分割：
library(caret) set.seed(998) inTraining <- createDataPartition(Sonar$Class, p = .75, list = FALSE) training <- Sonar[ inTraining,] #训练集 testing <- Sonar[-inTraining,] #测试集
模型拟合：
fitControl <- trainControl(## 10折交叉验证
method = "repeatedcv",
number = 10,
## 重复10次
repeats = 1)
set.seed(825)
gbmFit1 <- train(Class ~ ., data = training,
method = "gbm", # 助推树
trControl = fitControl,
verbose = FALSE)
gbmFit1
Stochastic Gradient Boosting
157 samples
60 predictor
2 classes: 'M', 'R'
No pre-processing
Resampling: Cross-Validated (10 fold, repeated 10 times)
Summary of sample sizes: 141, 142, 141, 142, 141, 142, ...
Resampling results across tuning parameters:
interaction.depth n.trees Accuracy Kappa
1 50 0.7935784 0.5797839
1 100 0.8171078 0.6290208
1 150 0.8219608 0.6383173
2 50 0.8041912 0.6027771
2 100 0.8296176 0.6544713
2 150 0.8283627 0.6520181
3 50 0.8110343 0.6170317
3 100 0.8301275 0.6551379
3 150 0.8310343 0.6577252
Tuning parameter 'shrinkage' was held constant at a value of 0.1
Tuning parameter 'n.minobsinnode' was held constant at a value of 10
Accuracy was used to select the optimal model using the largest value.
The final values used for the model were n.trees = 150, interaction.depth
= 3, shrinkage = 0.1 and n.minobsinnode = 10.
