spm (version 1.2.2)

RFcv: Cross validation, n-fold for random forest (RF)

Description

This function is a cross validation function for random forest.

Usage

RFcv(
  trainx,
  trainy,
  cv.fold = 10,
  mtry = if (!is.null(trainy) && !is.factor(trainy)) max(floor(ncol(trainx)/3), 1) else
    floor(sqrt(ncol(trainx))),
  ntree = 500,
  predacc = "ALL",
  ...
)

Arguments

trainx

a dataframe or matrix contains columns of predictor variables.

trainy

a vector of response, must have length equal to the number of rows in trainx.

cv.fold

integer; number of folds in the cross-validation. if > 1, then apply n-fold cross validation; the default is 10, i.e., 10-fold cross validation that is recommended.

mtry

a function of number of remaining predictor variables to use as the mtry parameter in the randomForest call.

ntree

number of trees to grow. This should not be set to too small a number, to ensure that every input row gets predicted at least a few times. By default, 500 is used.

predacc

can be either "VEcv" for vecv or "ALL" for all measures in function pred.acc.

...

other arguments passed on to randomForest.

Value

A list with the following components: for numerical data: me, rme, mae, rmae, mse, rmse, rrmse, vecv and e1; or vecv. for categorical data: correct classification rate (ccr), kappa (kappa), sensitivity (sens), specificity (spec) and true skill statistic (tss)

References

Li, J., J. Siwabessy, M. Tran, Z. Huang, and A. Heap. 2013. Predicting Seabed Hardness Using Random Forest in R. Pages 299-329 in Y. Zhao and Y. Cen, editors. Data Mining Applications with R. Elsevier.

Li, J. 2013. Predicting the spatial distribution of seabed gravel content using random forest, spatial interpolation methods and their hybrid methods. Pages 394-400 The International Congress on Modelling and Simulation (MODSIM) 2013, Adelaide.

Liaw, A. and M. Wiener (2002). Classification and Regression by randomForest. R News 2(3), 18-22.

Examples

# NOT RUN {
data(hard)
data(petrel)

rfcv1 <- RFcv(petrel[, c(1,2, 6:9)], petrel[, 5], predacc = "ALL")
rfcv1

n <- 20 # number of iterations, 60 to 100 is recommended.
VEcv <- NULL
for (i in 1:n) {
rfcv1 <- RFcv(petrel[, c(1,2,6:9)], petrel[, 5], predacc = "VEcv")
VEcv [i] <- rfcv1
}
plot(VEcv ~ c(1:n), xlab = "Iteration for RF", ylab = "VEcv (%)")
points(cumsum(VEcv) / c(1:n) ~ c(1:n), col = 2)
abline(h = mean(VEcv), col = 'blue', lwd = 2)

n <- 20 # number of iterations, 60 to 100 is recommended.
measures <- NULL
for (i in 1:n) {
rfcv1 <- RFcv(hard[, c(4:6)], hard[, 17])
measures <- rbind(measures, rfcv1$ccr) # for kappa, replace ccr with kappa
}
plot(measures ~ c(1:n), xlab = "Iteration for RF", ylab = "Correct
classification rate  (%)")
points(cumsum(measures) / c(1:n) ~ c(1:n), col = 2)
abline(h = mean(measures), col = 'blue', lwd = 2)
# }
# NOT RUN {
# }