R：随机林中的mtry调整错误（回归）

我有以下代码来调整随机森林回归模型的mtry超参数：

set.seed(42)

mtry <- 1:10

# Define train control
trControl <- trainControl(method = "cv",
                          number = 10,
                          search = "grid")

for (i in mtry) {
  rf_random <- train(Price.Gas~., data=data_train,
                 method = "rf",
                 mtry = i,
                 metric = "RMSE",
                 trControl = trControl)
}

---

我怎样才能测试不同的mtry值呢？

默认情况下，插入符号会在网格上调整mtry，请参见，这样您就不需要使用循环，而是在

tuneGrid=

中定义它：

library(caret)
set.seed(42)

data_train = data.frame(Price.Gas = rnorm(100),matrix(rnorm(1000),ncol=10))

trControl <- trainControl(method = "cv",number = 10)

rf_random <- train(Price.Gas~., data=data_train,
                   method = "rf",
                   tuneGrid = data.frame(mtry = 1:10),
                   metric = "RMSE",
                   trControl = trControl)

Random Forest 

100 samples
 10 predictor

No pre-processing
Resampling: Cross-Validated (10 fold) 
Summary of sample sizes: 89, 90, 91, 89, 91, 90, ... 
Resampling results across tuning parameters:

  mtry  RMSE       Rsquared   MAE      
   1    0.8556649  0.2122988  0.6921878
   2    0.8458829  0.2102749  0.6808978
   3    0.8518204  0.1975061  0.6909111
   4    0.8451160  0.1918390  0.6871511
   5    0.8386129  0.2037676  0.6808157
   6    0.8476718  0.1949056  0.6889514
   7    0.8434816  0.2082844  0.6833892
   8    0.8447137  0.1979602  0.6860908
   9    0.8419739  0.1960369  0.6825207
  10    0.8533284  0.1876459  0.6892574

RMSE was used to select the optimal model using the smallest value.
The final value used for the model was mtry = 5.

库（插入符号）
种子（42）
数据列=数据帧（Price.Gas=rnorm（100），矩阵（rnorm（1000），ncol=10））
trControl我收到此错误无效mtry:重置为在有效范围内。我想我选择了错误的mtry范围。我如何保证在不需要自己提供mtry值的情况下不会出现此错误？有什么经验法则吗？例如，您可以不指定
tuneGrid=
，而是指定
tuneLength=5
。插入符号将为您选择。我想你的mtry不能超过列数
library(caret)
set.seed(42)

data_train = data.frame(Price.Gas = rnorm(100),matrix(rnorm(1000),ncol=10))

trControl <- trainControl(method = "cv",number = 10)

rf_random <- train(Price.Gas~., data=data_train,
                   method = "rf",
                   tuneGrid = data.frame(mtry = 1:10),
                   metric = "RMSE",
                   trControl = trControl)

Random Forest 

100 samples
 10 predictor

No pre-processing
Resampling: Cross-Validated (10 fold) 
Summary of sample sizes: 89, 90, 91, 89, 91, 90, ... 
Resampling results across tuning parameters:

  mtry  RMSE       Rsquared   MAE      
   1    0.8556649  0.2122988  0.6921878
   2    0.8458829  0.2102749  0.6808978
   3    0.8518204  0.1975061  0.6909111
   4    0.8451160  0.1918390  0.6871511
   5    0.8386129  0.2037676  0.6808157
   6    0.8476718  0.1949056  0.6889514
   7    0.8434816  0.2082844  0.6833892
   8    0.8447137  0.1979602  0.6860908
   9    0.8419739  0.1960369  0.6825207
  10    0.8533284  0.1876459  0.6892574

RMSE was used to select the optimal model using the smallest value.
The final value used for the model was mtry = 5.