Regression.CV.Batch.Fit: CV Batch Training and Diagnostics of Regression Base Learners

Description

CV Batch Training and Diagnostics of Regression Base Learners.

Usage

Regression.CV.Batch.Fit(instance.list, formula, data
  , ncores = 1, filemethod = FALSE, print.level = 1
  , preschedule = TRUE
  , schedule.method = c("random", "as.is", "task.length")
  , task.length)
# S3 method for Regression.CV.Batch.FitObj
predict(object, ..., ncores=1
  , preschedule = TRUE)
# S3 method for Regression.CV.Batch.FitObj
plot(x, errfun=rmse.error, ylim.adj = NULL, ...)

Value

Function Regression.CV.Batch.Fit produces an object of class Regression.CV.Batch.FitObj. The predict method produces a matrix, whose columns each represent training-set predictions from one of the batch of base learners (in CV fashion).

Arguments

instance.list: An object of class Instance.List, containing all combinations of base learner configurations and data partitions to perform CV batch training.
formula: Formula object expressing response variable and covariates.
data: Data frame expressing response variable and covariates.
ncores: Number of cores in parallel training.
filemethod: Boolean flag, indicating whether to save estimation objects to file or not.
print.level: Verbosity level.
preschedule: Boolean flag, indicating whether parallel jobs must be scheduled statically (TRUE) or dynamically (FALSE).
schedule.method: Method used for scheduling tasks across threads. In as.is, tasks are distributed in round-robin fashion. In random, tasks are randomly shuffled before round-robin distribution. In task.length, estimated task execution times are used to allocate them to threads to maximize load balance.
task.length: Estimation task execution times, to be used for loading balancing during parallel execution.
object: Output of Regression.CV.Batch.Fit, object of class Regression.CV.Batch.FitObj.
...: Arguments passed from/to other functions.
x: Object of class Regression.CV.Batch.FitObj, to creates a plot from.
errfun: Error function used in generating plot.
ylim.adj: Optional numeric argument to use for adjusting the range of y-axis.

Author

Alireza S. Mahani, Mansour T.A. Sharabiani

Examples

Run this code

data(servo)
myformula <- class~motor+screw+pgain+vgain

perc.train <- 0.7
index.train <- sample(1:nrow(servo)
  , size = round(perc.train*nrow(servo)))
data.train <- servo[index.train,]
data.predict <- servo[-index.train,]

parts <- generate.partitions(1, nrow(data.train))
myconfigs <- make.configs("knn"
  , config.df = expand.grid(kernel = "rectangular", k = c(5, 10)))
instances <- make.instances(myconfigs, parts)

ret <- Regression.CV.Batch.Fit(instances, myformula, data.train)
newpred <- predict(ret, data.predict)