summary.cv: Summarize cross-validation results

In cvTools: Cross-validation tools for regression models

Description

Produce a summary of results from (repeated) K-fold cross-validation.

Usage

  ## S3 method for class 'cv'
 summary(object, ...)

  ## S3 method for class 'cvSelect'
 summary(object, ...)

  ## S3 method for class 'cvTuning'
 summary(object, ...)

Arguments

object	an object inheriting from class "cv" or "cvSelect" that contains cross-validation results (note that the latter includes objects of class "cvTuning").
...	currently ignored.

Value

An object of class "summary.cv", "summary.cvSelect" or "summary.cvTuning", depending on the class of object.

Author(s)

Andreas Alfons

See Also

cvFit, cvSelect, cvTuning, summary

Examples

library("robustbase")
data("coleman")
set.seed(1234)  # set seed for reproducibility

## set up folds for cross-validation
folds <- cvFolds(nrow(coleman), K = 5, R = 10)


## compare raw and reweighted LTS estimators for 
## 50% and 75% subsets

# 50% subsets
fitLts50 <- ltsReg(Y ~ ., data = coleman, alpha = 0.5)
cvFitLts50 <- cvLts(fitLts50, cost = rtmspe, folds = folds, 
    fit = "both", trim = 0.1)

# 75% subsets
fitLts75 <- ltsReg(Y ~ ., data = coleman, alpha = 0.75)
cvFitLts75 <- cvLts(fitLts75, cost = rtmspe, folds = folds, 
    fit = "both", trim = 0.1)

# combine results into one object
cvFitsLts <- cvSelect("0.5" = cvFitLts50, "0.75" = cvFitLts75)
cvFitsLts

# summary of the results with the 50% subsets
summary(cvFitLts50)
# summary of the combined results
summary(cvFitsLts)


## evaluate MM regression models tuned for 
## 80%, 85%, 90% and 95% efficiency
tuning <- list(tuning.psi=c(3.14, 3.44, 3.88, 4.68))

# set up function call
call <- call("lmrob", formula = Y ~ .)
# perform cross-validation
cvFitsLmrob <- cvTuning(call, data = coleman, 
    y = coleman$Y, tuning = tuning, cost = rtmspe, 
    folds = folds, costArgs = list(trim = 0.1))
cvFitsLmrob

# summary of results
summary(cvFitsLmrob)

Example output

Loading required package: lattice
Loading required package: robustbase

5-fold CV results:
   Fit reweighted      raw
1  0.5   1.140772 1.511817
2 0.75   0.963192 1.165930

Best model:
reweighted        raw 
    "0.75"     "0.75" 
5-fold CV results:
        reweighted      raw
Min.     0.8426695 1.121681
1st Qu.  0.9832825 1.476071
Median   1.1400160 1.502916
Mean     1.1407723 1.511817
3rd Qu.  1.2835384 1.623199
Max.     1.5522526 1.875849

5-fold CV results:
   Fit reweighted.Min. reweighted.1st Qu. reweighted.Median reweighted.Mean
1  0.5       0.8426695          0.9832825         1.1400160       1.1407723
2 0.75       0.8360649          0.8979938         0.9829863       0.9631920
  reweighted.3rd Qu. reweighted.Max. raw.Min. raw.1st Qu. raw.Median raw.Mean
1          1.2835384       1.5522526 1.121681    1.476071   1.502916 1.511817
2          1.0241433       1.0850610 1.046983    1.097745   1.145762 1.165930
  raw.3rd Qu. raw.Max.
1    1.623199 1.875849
2    1.234328 1.298372

Best model:
reweighted        raw 
    "0.75"     "0.75" 
Warning message:
In lf.cov(init, x = x) :
  .vcov.avar1: negative diag(<vcov>) fixed up; consider 'cov=".vcov.w."' instead

5-fold CV results:
  tuning.psi        CV
1       3.14 0.9683601
2       3.44 0.9654708
3       3.88 0.9681099
4       4.68 0.9865325

Optimal tuning parameter:
   tuning.psi
CV       3.44

5-fold CV results:
  tuning.psi   CV.Min. CV.1st Qu. CV.Median   CV.Mean CV.3rd Qu.   CV.Max.
1       3.14 0.8387515  0.9198769 0.9743848 0.9683601  1.0278803 1.0829762
2       3.44 0.8375457  0.9203998 0.9701659 0.9654708  1.0197471 1.0855025
3       3.88 0.8375048  0.9220203 0.9713238 0.9681099  1.0395408 1.0982037
4       4.68 0.8087384  0.9208760 0.9892362 0.9865325  1.0601024 1.1479685

Optimal tuning parameter:
   tuning.psi
CV       3.44

cvTools documentation built on May 2, 2019, 12:53 a.m.