Version:	0.1-5
Date:	2023-04-11
Title:	Stability Assessment of Statistical Learning Methods
Description:	Graphical and computational methods that can be used to assess the stability of results from supervised statistical learning.
Depends:	R (≥ 3.0.0)
Imports:	graphics, methods, MASS, e1071, partykit, party, randomForest, ranger
Suggests:	utils, Formula, nnet, rpart, evtree, rchallenge, knitr, rmarkdown
VignetteBuilder:	knitr
License:	GPL-2 | GPL-3
Encoding:	UTF-8
NeedsCompilation:	no
Packaged:	2023-04-12 23:53:11 UTC; zeileis
Author:	Michel Philipp [aut], Carolin Strobl [aut], Achim Zeileis [aut, cre], Thomas Rusch [aut], Kurt Hornik [aut], Lennart Schneider [aut]
Maintainer:	Achim Zeileis <Achim.Zeileis@R-project.org>
Repository:	CRAN
Date/Publication:	2023-04-13 12:22:20 UTC

List of Predefined Learners for Assessing Stability

Description

The list contains details about several predefined learners that are required to assess the stability of results from statistical learning.

Usage

LearnerList

Details

Currently implemented learners are:

ctree

conditional inference trees using ctree from partykit.

rpart

recursive partitioning using rpart from rpart.

J48

recursive partitioning using J48 from RWeka.

C5.0

recursive partitioning using C5.0 from C50.

tree

recursive partitioning using tree from tree.

lda

linear discriminant analysis using lda from MASS.

lm

linear models using lm from stats.

glm

generalized linear models using glm from stats.

Users can add new learners to LearnerList for the current R session, see addLearner.

See Also

addLearner

Prediction Accuracy from Stability Assessment Results

Description

Function to compute the prediction accuracy from an object of class "stablelearner" or "stablelearnerList" as a parallel to the similarity values estimated by stability in each iteration of the stability assessment procedure.

Usage

  accuracy(x, measure = "kappa", na.action = na.exclude, 
	   applyfun = NULL, cores = NULL) 

Arguments

Details

This function can be used to compute prediction accuracy after the stability was estimated using stability.

Value

A matrix of size 2*B times length(measure) containing prediction accuracy values of the learners trained during the stability assessment procedure.

See Also

stability

Examples

library("partykit")
res <- ctree(Species ~ ., data = iris)
stab <- stability(res)
accuracy(stab)

Add Learners to LearnerList

Description

The function can be used to add new learner to LearnerList in the current R session.

Usage

addLearner(x)

Arguments

Details

The function can be used to add new learners to LearnerList in the current R session. The function expects a list of four elements including the name of the learners object class, the name of the package where the class and the fitting method is implemented, the name of the method and a prediction function that predicts class probabilities (in the classification case) or numeric values (in the regression case) and takes the arguments x (the fitted model object), newdata a data.frame containing the predictions of the observations in the evaluation sample and yclass a character string specifying the type of the response variable ("numeric", "factor", etc.). The elements in the list should be named class, package, method and predfun.

See Also

LearnerList

Examples

newlearner <- list(
  class   = "svm",
  package = "e1071",
  method  = "Support Vector Machine",
  predict = function(x, newdata, yclass = NULL) {
    if(match(yclass, c("ordered", "factor"))) {
      attr(predict(x, newdata = newdata, probability = TRUE), "probabilities")
    } else {
      predict(x, newdata = newdata)
    }
  })

addLearner(newlearner)

Sampler Infrastructure for Stability Assessment

Description

Sampler objects that provide objects with functionality used by stabletree to generate resampled datasets.

Usage

  bootstrap(B = 500, v = 1)
  subsampling(B = 500, v = 0.632)
  samplesplitting(k = 5)
  jackknife(d = 1, maxrep = 5000)
  splithalf(B = 500)

Arguments

Details

The sampler functions provide objects that include functionality to generate resampled datasets used by stabletree.

The bootstrap function provides an object that can be used to generate B bootstrap samples by sampling from n observations with replacement.

The subsampling function provides an object that can be used to generate B subsamples by sampling from floor(v*n) observations without replacement.

The samplesplitting function provides an object that can be used to generate k-folds from n observations.

The jackknife function provides an object that can be used to generate all datasets necessary to perform leave-k-out jackknife sampling from n observations. The number of datasets is limited by maxrep to prevent unintended CPU or memory overload by accidently choosing too large values for k.

The splithalf function provides an object that can be used to generate B subsamples by sampling from floor(0.5*n) observations without replacement. When used to implement the "splithalf" resampling strategy for measuring the stability of a result via the stability function, the matrix containing the complement learning samples is generated automatically by stability.

See Also

stabletree, stability

Examples

set.seed(0)

## bootstrap sampler
s <- bootstrap(3)
s$sampler(10)

## subsampling
s <- subsampling(3, v = 0.6)
s$sampler(10)

## 5-fold sample-splitting
s <- samplesplitting(5)
s$sampler(10)

## jackknife
s <- jackknife(d = 1)
s$sampler(10)

## splithaf
s <- splithalf(3)
s$sampler(10)

Illustrate Results from Stability Assessment

Description

Illustrates the results from stability assessments performed by stability using boxplots.

Usage

  ## S3 method for class 'stablelearnerList'
boxplot(x, ..., main = NULL, xlab = NULL, ylab = NULL, reverse = TRUE)
  ## S3 method for class 'stablelearner'
boxplot(x, ...)

Arguments

See Also

stability, summary.stablelearnerList

Examples

library("partykit")
r1 <- ctree(Species ~ ., data = iris)

library("rpart")
r2 <- rpart(Species ~ ., data = iris)

stab <- stability(r1, r2, names = c("ctree", "rpart"))
boxplot(stab)

Data-Ggnerating Function for Two-Class Problem

Description

Data-generating function to generate artificial data sets of a classification problem with two response classes, denoted as "A" and "B".

Usage

  dgp_twoclass(n = 100, p = 4, noise = 16, rho = 0, 
    b0 = 0, b = rep(1, p), fx = identity)

Arguments

Value

A data.frame including a column denoted as class that is a factor with two levels "A" and "B". All other columns represent the predictor variables (signal predictors followed by noise predictors) and are named by "x1", "x2", etc..

See Also

stability

Examples

dgp_twoclass(n = 200, p = 6, noise = 4)

Get Learner Details from LearnerList

Description

Function to get information available about a specific learner in LearnerList of the current R session.

Usage

getLearner(x)

Arguments

Details

The function returns the entry in LearnerList found for the class of the object submitted to the function.

See Also

LearnerList, addLearner

Examples

library("partykit")
m <- ctree(Species ~ ., data = iris)
getLearner(m)

Visualizing Tree Stability Assessments

Description

Visualizations of tree stability assessments carried out via stabletree.

Usage

## S3 method for class 'stabletree'
plot(x, select = order(colMeans(x$vs), decreasing = TRUE), 
  type.breaks = "levels", col.breaks = "red", lty.breaks = "dashed",
  cex.breaks = 0.7,  col.main = c("black", "gray50"), main.uline = TRUE,
  args.numeric = NULL, args.factor = NULL, args.ordered = NULL, main = NULL,
  original = TRUE, ...)

## S3 method for class 'stabletree'
barplot(height, main = "Variable selection frequencies",
  xlab = "", ylab = "", horiz = FALSE, col = gray.colors(2),
  names.arg = NULL, names.uline = TRUE, names.diag = TRUE,
  cex.names = 0.9, 
  ylim = if (horiz) NULL else c(0, 100), xlim = if (horiz) c(0, 100) else NULL,
  original = TRUE, ...)

## S3 method for class 'stabletree'
image(x, main = "Variable selections",
  ylab = "Repetitions", xlab = "", col = gray.colors(2),
  names.arg = NULL, names.uline = TRUE, names.diag = TRUE, 
  cex.names = 0.9, xaxs = "i", yaxs = "i",
  col.tree = 2, lty.tree = 2, xlim = c(0, length(x$vs0)), ylim = c(0, x$B),
  original = TRUE, ...)

Arguments

Details

• plot visualizes the variability of the cutpoints.

• barplot visualizes the variable selection frequency.

• image visualizes the combinations of variables selected.

See Also

stabletree

Examples

## build a tree
library("partykit")
m <- ctree(Species ~ ., data = iris)
plot(m)

## investigate stability
set.seed(0)
s <- stabletree(m, B = 500)

## show variable selection proportions
## with different labels and different ordering
barplot(s)
barplot(s, cex.names = 0.8)
barplot(s, names.diag = FALSE)
barplot(s, names.arg = c("a", "b", "c", "d"))
barplot(s, names.uline = FALSE)
barplot(s, col = c("lightgreen", "darkred"))
barplot(s, horiz = TRUE)

## illustrate variable selections of replications
## with different labels and different ordering
image(s)
image(s, cex.names = 0.8)
image(s, names.diag = FALSE)
image(s, names.arg = c("a", "b", "c", "d"))
image(s, names.uline = FALSE)
image(s, col = c("lightgreen", "darkred"))

## graphical cutpoint analysis, selecting variable by number and name
## with different numerical of break points
plot(s)
plot(s, select = 3)
plot(s, select = "Petal.Width")
plot(s, args.numeric = list(breaks = 40))

# change labels of splits in complete data tree
plot(s, select = 3, type.breaks = "levels")
plot(s, select = 3, type.breaks = "nodeids")
plot(s, select = 3, type.breaks = "breaks")
plot(s, select = 3, type.breaks = "none")

Similarity Measure Infrastructure for Stability Assessment with Ordinal Responses

Description

Functions that provide objects with functionality used by stability to measure the similarity between the predictions of two results in classification problems.

Usage

  clagree()
  ckappa()

  bdist()
  tvdist()
  hdist()
  jsdiv(base = 2)

Arguments

Details

The similarity measure functions provide objects that include functionality used by stability to measure the similarity between the probability predictions of two results in classification problems.

The clagree and ckappa functions provide an object that can be used to assess the similarity based on the predicted classes of two results. The predicted classes are selected by the class with the highest probability.

The bdist (Bhattacharayya distance), tvdist (Total variation distance), hdist (Hellinger distance) and jsdist (Jenson-Shannon divergence) functions provide an object that can be used to assess the similarity based on the predicted class probabilities of two results.

See Also

stability

Examples

set.seed(0)

## build trees
library("partykit")
m1 <- ctree(Species ~ ., data = iris[sample(1:nrow(iris), replace = TRUE),])
m2 <- ctree(Species ~ ., data = iris[sample(1:nrow(iris), replace = TRUE),])

p1 <- predict(m1, type = "prob")
p2 <- predict(m2, type = "prob")

## class agreement
m <- clagree()
m$measure(p1, p2)

## cohen's kappa
m <- ckappa()
m$measure(p1, p2)

## bhattacharayya distance
m <- bdist()
m$measure(p1, p2)

## total variation distance
m <- tvdist()
m$measure(p1, p2)

## hellinger distance
m <- hdist()
m$measure(p1, p2)

## jenson-shannon divergence
m <- jsdiv()
m$measure(p1, p2)

## jenson-shannon divergence (base = exp(1))
m <- jsdiv(base = exp(1))
m$measure(p1, p2)

Similarity Measure Infrastructure for Stability Assessment with Numerical Responses

Description

Functions that provide objects with functionality used by stability to measure the similarity between numeric predictions of two results in regression problems.

Usage

  edist()
  msdist()
  rmsdist()
  madist()
  qadist(p = 0.95)
  
  cprob(kappa = 0.1)
  rbfkernel()
  tanimoto()
  cosine()
  
  ccc()
  pcc()

Arguments

Details

The similarity measure functions provide objects that include functionality used by stability to measure the similarity between numeric predictions of two results in regression problems.

The edist (euclidean distance), msdist (mean squared distance), rmsdist (root mean squared distance), madist (mean absolute distance) and qadist (quantile of absolute distance) functions implement scale-variant distance measures that are unbounded.

The cprob (coverage probability), rbfkernel (gaussian radial basis function kernel), tanimoto (tanimoto coefficient) and cosine (cosine similarity) functions implement scale-variant distance measures that are bounded.

The ccc (concordance correlation coefficient) and pcc (pearson correlation coefficient) functions implement scale-invariant distance measures that are bounded between 0 and 1.

See Also

stability

Examples

set.seed(0)

library("partykit")

airq <- subset(airquality, !is.na(Ozone))
m1 <- ctree(Ozone ~ ., data = airq[sample(1:nrow(airq), replace = TRUE),])
m2 <- ctree(Ozone ~ ., data = airq[sample(1:nrow(airq), replace = TRUE),])

p1 <- predict(m1)
p2 <- predict(m2)

## euclidean distance
m <- edist()
m$measure(p1, p2)

## mean squared distance
m <- msdist()
m$measure(p1, p2)

## root mean squared distance
m <- rmsdist()
m$measure(p1, p2)

## mean absolute istance
m <- madist()
m$measure(p1, p2)

## quantile of absolute distance
m <- qadist()
m$measure(p1, p2)

## coverage probability
m <- cprob()
m$measure(p1, p2)

## gaussian radial basis function kernel
m <- rbfkernel()
m$measure(p1, p2)

## tanimoto coefficient
m <- tanimoto()
m$measure(p1, p2)

## cosine similarity
m <- cosine()
m$measure(p1, p2)

## concordance correlation coefficient
m <- ccc()
m$measure(p1, p2)

## pearson correlation coefficient
m <- pcc()
m$measure(p1, p2)

Extracting Similarity Values

Description

Extract similarity values from object returned by stability for further illustration or analysis.

Usage

  similarity_values(x, reverse = TRUE)

Arguments

Value

A numeric array of dimension 3 containing similarity values. The dimensions represent repetitions, results (fitted model objects) and similarity measures.

See Also

stability, summary.stablelearnerList

Examples

library("partykit")
res <- ctree(Species ~ ., data = iris)
stab <- stability(res)
similarity_values(stab)

Control for Supervised Stability Assessments

Description

Various parameters that control aspects of the stability assessment performed via stability.

Usage

  stab_control(B = 500, measure = list(tvdist, ccc), sampler = "bootstrap", 
    evaluate = "OOB", holdout = 0.25, seed = NULL, na.action = na.exclude,
    savepred = TRUE, silent = TRUE, ...)

Arguments

Details

With the argument measure one or more measures can be defined that are used to assess the stability of a result from supervised statistical learning by stability. Predefined similarity measures for the regression and the classification case are listed in similarity_measures_classification and similarity_measures_regression.

Users can define their own similarity functions f(p1, p2) that must return a single numeric value for the similarity between two results trained on resampled data sets. Such a function must take the arguments p1 and p2. In the classification case, p1 and p2 are probability matrices of size m * K, where m is the number of predicted observations (size of the evaluation sample) and K is the number of classes. In the regression case, p1 and p2 are numeric vectors of length m.

A different way to implement new similarity functions for the current R session is to define a similarity measure generator function, which is a function without arguments that generates a list of five elements including the name of the similarity measure, the function to compute the similarity between the predictions as described above, a vector of character values specifying the response types for which the similarity measure can be used, a list containing two numeric elements lower and upper that specify the range of values of the similarity measure and the function to invert (or reverse) the similarity values such that higher values indicate higher stability. The latter can be set to NULL, if higher similarity values already indicate higher stability. Those elements should be named name, measure, classes, range and reverse.

The argument evaluate can be used to specify the evaluation strategy. If set to "ALL", all observations in the original data set are used for evaluation. If set to "OOB", only the pairwise out-of-bag observations are used for evaluation within each repetition. If set to "OOS", a fraction (defined by holdout) of the observations in the original data set are randomly sampled and used for evaluation, but not for training, over all repetitions.

The argument seed can be used to make similarity assessments comparable when comparing the stability of different results that were trained on the same data set. By default, seed is set to NULL and the learning samples are sampled independently for each fitted model object passed to stability. If seed is set to a specific number, the seed will be set for each fitted model object before the learning samples are generated using "L'Ecuyer-CMRG" (see set.seed) which guarantees identical learning samples for each stability assessment and, thus, comparability of the stability assessments between the results.

See Also

stability

Examples

library("partykit")
res <- ctree(Species ~ ., data = iris)

## less repetitions
stability(res, control = stab_control(B = 100))



## Not run: 

## change similarity measure
stability(res, control = stab_control(measure = list(bdist)))

## change evaluation strategy
stability(res, control = stab_control(evaluate = "ALL"))
stability(res, control = stab_control(evaluate = "OOS"))

## change resampling strategy to subsampling
stability(res, control = stab_control(sampler = subsampling))
stability(res, control = stab_control(sampler = subsampling, evaluate = "ALL"))
stability(res, control = stab_control(sampler = subsampling, evaluate = "OOS"))

## change resampling strategy to splithalf
stability(res, control = stab_control(sampler = splithalf, evaluate = "ALL"))
stability(res, control = stab_control(sampler = splithalf, evaluate = "OOS"))


## End(Not run)

Stability Assessment for Results from Supervised Statistical Learning

Description

Stability assessment of results from supervised statistical learning (i.e., recursive partitioning, support vector machines, neural networks, etc.). The procedure involves the pairwise comparison of results generated from learning samples randomly drawn from the original data set or directly from the data-generating process (if available).

Usage

  stability(x, ..., data = NULL, control = stab_control(), weights = NULL, 
    applyfun = NULL, cores = NULL, names = NULL)

Arguments

Details

Assesses the (overall) stability of a result from supervised statistical learning by quantifying the similarity of realizations from the distribution of possible results (given the algorithm, the formulated model, the data-generating process, the sample size, etc.). The stability distribution is estimated by repeatedly assessing the similarity between the results generated by training the algorithm on two different learning samples, by means of a similarity metric. The learning samples are generated by sampling from the learning data or the data-generating process in case of a simulation study. For more details, see Philipp et al. (2018).

Value

For a single fitted model object, stability returns an object of class "stablelearner" with the following components:

For several fitted model objects, stability returns an object of class "stablelearnerList" which is a list of objects of class "stablelearner".

References

Philipp M, Rusch T, Hornik K, Strobl C (2018). “Measuring the Stability of Results from Supervised Statistical Learning”. Journal of Computational and Graphical Statistics, 27(4), 685–700. doi:10.1080/10618600.2018.1473779

See Also

boxplot.stablelearnerList, summary.stablelearner

Examples

## assessing the stability of a single result
library("partykit")
r1 <- ctree(Species ~ ., data = iris)
stab <- stability(r1)
summary(stab)

## assessing the stability of several results
library("rpart")
r2 <- rpart(Species ~ ., data = iris)
stab <- stability(r1, r2, control = stab_control(seed = 0))
summary(stab, names = c("ctree", "rpart"))

## using case-weights instead of resampling
stability(r1, weights = TRUE)

## using self-defined case-weights
n <- nrow(iris)
B <- 500
w <- array(sample(c(0, 1), size = n * B * 3, replace = TRUE), dim = c(n, B, 3))
stability(r1, weights = w)

## assessing stability for a given data-generating process
my_dgp <- function() dgp_twoclass(n = 100, p = 2, noise = 4, rho = 0.2)
res <- ctree(class ~ ., data = my_dgp())
stability(res, data = my_dgp)

Stability Assessment for Tree Learners

Description

Stability assessment of variable and cutpoint selection in tree learners (i.e., recursive partitioning). By refitting trees to resampled versions of the learning data, the stability of the trees structure is assessed and can be summarized and visualized.

Usage

  stabletree(x, data = NULL, sampler = subsampling, weights = NULL,
    applyfun = NULL, cores = NULL, savetrees = FALSE, ...)

Arguments

Details

The function stabletree assesses the stability of tree learners (i.e., recursive partitioning methods) by refitting the tree to resampled versions of the learning data. By default, if data = NULL, the fitting call is extracted by getCall to infer the learning data. Subsequently, the sampler generates B resampled versions of the learning data, the tree is regrown with update, and (if necessary) coerced by as.party. For each of the resampled trees it is queried and stored which variables are selected for splitting and what the selected cutpoints are.

The resulting object of class "stabletree" comes with a set of standard methods to generic functions including print, summary for numerical summaries and plot, barplot, and image for graphical representations. See plot.stabletree for more details. In most methods, the argument original can be set to TRUE or FALSE, turning highlighting of the original tree information on and off.

Value

stabletree returns an object of class "stabletree" which is a list with the following components:

References

Hothorn T, Zeileis A (2015). partykit: A Modular Toolkit for Recursive Partytioning in R. Journal of Machine Learning Research, 16(118), 3905–3909.

Philipp M, Zeileis A, Strobl C (2016). “A Toolkit for Stability Assessment of Tree-Based Learners”. In A. Colubi, A. Blanco, and C. Gatu (Eds.), Proceedings of COMPSTAT 2016 – 22nd International Conference on Computational Statistics (pp. 315–325). The International Statistical Institute/International Association for Statistical Computing. Preprint available at https://EconPapers.RePEc.org/RePEc:inn:wpaper:2016-11

See Also

plot.stabletree, as.stabletree, as.party

Examples

## build a simple tree
library("partykit")
m <- ctree(Species ~ ., data = iris)
plot(m)

## investigate stability
set.seed(0)
s <- stabletree(m, B = 500)
print(s)

## variable selection statistics
summary(s)

## show variable selection proportions
barplot(s)

## illustrate variable selections of replications
image(s)

## graphical cutpoint analysis
plot(s)

Coercion Functions

Description

Functions coercing various forest objects to objects of class "stabletree".

Usage

  as.stabletree(x, ...)
  ## S3 method for class 'randomForest'
as.stabletree(x, applyfun = NULL, cores = NULL, ...)
  ## S3 method for class 'RandomForest'
as.stabletree(x, applyfun = NULL, cores = NULL, ...)
  ## S3 method for class 'cforest'
as.stabletree(x, applyfun = NULL, cores = NULL, savetrees = FALSE, ...)
  ## S3 method for class 'ranger'
as.stabletree(x, applyfun = NULL, cores = NULL, ...)

Arguments

Details

Random forests fitted using randomForest, cforest, cforest or ranger are coerced to "stabletree" objects.

Note that when plotting a randomForest or ranger, the gray areas of levels of a nominal variable do not mimic exactly the same behavior as for classical "stabletree" objects, due to randomForest and ranger, not storing any information whether any individuals were left fulfilling the splitting criterion in the subsample. Therefore, gray areas only indicate that this level of this variable has already been used in a split before in such a way that it could not be used for any further splits.

For ranger, interaction terms are (currently) not supported.

Value

as.stabletree returns an object of class "stabletree" which is a list with the following components:

See Also

stabletree, plot.stabletree

Examples

## build a randomForest using randomForest
library(randomForest)
set.seed(1)
rf <- randomForest(Species ~ ., data = iris)

## coerce to a stabletree
srf <- as.stabletree(rf)
print(srf)
summary(srf, original = FALSE) # there is no original tree
barplot(srf)
image(srf)
plot(srf)

## build a RandomForest using party
library("party")
set.seed(2)
cf_party <- cforest(Species ~ ., data = iris,
  control = cforest_unbiased(mtry = 2))

## coerce to a stabletree
scf_party <- as.stabletree(cf_party)
print(scf_party)
summary(scf_party, original = FALSE)
barplot(scf_party)
image(scf_party)
plot(scf_party)

## build a cforest using partykit
library("partykit")
set.seed(3)
cf_partykit <- cforest(Species ~ ., data = iris)

## coerce to a stabletree
scf_partykit <- as.stabletree(cf_partykit)
print(scf_partykit)
summary(scf_partykit, original = FALSE)
barplot(scf_partykit)
image(scf_partykit)
plot(scf_partykit)

## build a random forest using ranger
library("ranger")
set.seed(4)
rf_ranger <- ranger(Species ~ ., data = iris)

## coerce to a stabletree
srf_ranger <- as.stabletree(rf_ranger)
print(srf_ranger)
summary(srf_ranger, original = FALSE)
barplot(srf_ranger)
image(srf_ranger)
plot(srf_ranger)

Summarize Results from Stability Assessment

Description

Summarizes and prints the results from stability assessments performed by stability.

Usage

## S3 method for class 'stablelearnerList'
summary(object, ..., reverse = TRUE, 
  probs = c(0.05, 0.25, 0.5, 0.75, 0.95), digits = 3, names = NULL)

Arguments

See Also

stability

Examples

library("partykit")

rval <- ctree(Species ~ ., data = iris)
stab <- stability(rval)

summary(stab)
summary(stab, reverse = FALSE)
summary(stab, probs = c(0.25, 0.5, 0.75))
summary(stab, names = "conditional inference tree")

Passengers and Crew on the RMS Titanic

Description

the titanic data is a complete list of passengers and crew members on the RMS Titanic. It includes a variable indicating whether a person did survive the sinking of the RMS Titanic on April 15, 1912.

Usage

data("titanic")

Format

A data frame containing 2207 observations on 11 variables.

name

a string with the name of the passenger.

gender

a factor with levels male and female.

age

a numeric value with the persons age on the day of the sinking. The age of babies (under 12 months) is given as a fraction of one year (1/month).

class

a factor specifying the class for passengers or the type of service aboard for crew members.

embarked

a factor with the persons place of of embarkment.

country

a factor with the persons home country.

ticketno

a numeric value specifying the persons ticket number (NA for crew members).

fare

a numeric value with the ticket price (NA for crew members, musicians and employees of the shipyard company).

sibsp

an ordered factor specifying the number if siblings/spouses aboard; adopted from Vanderbild data set (see below).

parch

an ordered factor specifying the number of parents/children aboard; adopted from Vanderbild data set (see below).

survived

a factor with two levels (no and yes) specifying whether the person has survived the sinking.

Details

The website https://www.encyclopedia-titanica.org/ offers detailed information about passengers and crew members on the RMS Titanic. According to the website 1317 passengers and 890 crew member were abord.

8 musicians and 9 employees of the shipyard company are listed as passengers, but travelled with a free ticket, which is why they have NA values in fare. In addition to that, fare is truely missing for a few regular passengers.

Source

The complete list of persons on the RMS titanic was downloaded from https://www.encyclopedia-titanica.org/ on April 5, 2016. The information given in sibsp and parch was adopoted from a data set obtained from https://hbiostat.org/data/.

References

https://www.encyclopedia-titanica.org/ and https://hbiostat.org/data/.

Examples

data("titanic", package = "stablelearner")
summary(titanic)

Tuning Wrapper Function

Description

Convenience function to train a method using different tuning parameters.

Usage

  tuner(method, tunerange, ...)

Arguments

Details

This function can be used to train any method using different values for its tuning parameter(s). The result can be passed directly to stability to compare the stability of results based on different values of the tuning parameter.

Value

A list that contains all fitted model objects.

Additional information about the range of values used for the tuning parameters is attached to the resulting object as an attribute.

See Also

stability

Examples

library("partykit")

## tuning cforest using different values of its tuning parameter mtry
r <- tuner("cforest", tunerange = list(mtry = 1:4), formula = Species ~ ., data = iris)

## assess stability (with B = 10 for illustration to avoid excessive computation times)
stability(r, control = stab_control(seed = 1234, B = 10))

## receive information about the range of tuning parameters
attr(r, "range")