Changelog

Source: NEWS.md

RFSurrogates 0.4.0

Added RandomForestSurrogates().

• This functions aims to replace the first section of the variable selection and relation functions by creating a single reusable object which contains the random forest ranger::ranger() model, as well as the trees list with layers and surrogates added.
• Returns a RandomForestSurrogates object, which serves as the base object for later analysis.
• Additional ... params are passed directly to ranger::ranger.
• s.pct is a helper for calculating the number of surrogates as a fraction of number of variables (Default: 0.01). s can be set to overwrite this default.
• mtry supports the following values:
  • One of the documented string values, which will cause the mtry passed to ranger::ranger() to be a function accepting the number of variables, and returning the specific transformation after flooring the result.
  • A function which takes the number of variables as its first and only param, and returns the value of mtry.
  • A numeric value for mtry.
  • The default is "^3/4".
• type also uses match.arg() and still defaults to "regression".
• num.threads defaults to 1.
• permutate will, if set to TRUE, apply random permutation to the data in each feature. (This is used in permutation importance approaches.)
• seed is now a strongly recommended optional parameter (issuing a warning whenever it is not set).
  • Setting seed will cause a call to set.seed() when permutating. It is also used as the seed param of the ranger::ranger() call.
  • Requiring seed as a function parameter is preferred because it does not rely on global, non-reproducible state of the random number generator, if it was not seeded immediately before the function call.
• The inner call to ranger::ranger() includes the following defaults:
  • keep.inbag = TRUE
  • respect.unordered.factors = "partition"
  • Data is passed as a data.frame with the special column y, and the optional special column status for survival forests.
    • x must not contain the column names y or status, as this may lead to unexpected behavior.
• In general, input parameters are more strictly validated.
• The function uses num.threads to also parallelize creating the list of trees with layers.

Added SurrogateMinimalDepth() as a replacement for var.select.smd(). It takes just a RandomForestSurrogates object as a param, and returns a SurrogateMinimalDepth list object.

Value:

• RFS: The RandomForestSurrogates object used.
• selected: The selected variables (previously var).
• depth: The surrogate minimal depth of each variable (previously info$depth).
• threshold: The threshold used to select variables (previously info$threshold).
• surrogates: Details on the average number of surrogates in the forest, and per layer.
  • average: Previously s$s.a.
  • layer: Previously s$s.l.

Added MeanAdjustedAgreement() as a replacement for var.relations(). It takes a RandomForestSurrogates object as a param, and returns a MeanAdjustedAgreement list object.

Value:

• RFS: The RandomForestSurrogates object used.
• relations: The matrix of mean adjusted agreements for the investigated variables (rows) and the possible candidates (columns) (previously surr.res).
• threshold: The threshold used to select related variables.
• related: A list of vectors for each investigated variable containing the related candidate variables (previously var).

Added MFI() as replacement for var.relations.mfi(). MFI() takes the same arguments as RandomForestSurrogates(), with the additional variables and candidates params. Perform variable selection using MutualForestImpactVariableSelection().

Added MutualImpurityReduction() as a replacement for var.select.mir(), where variable selection is performed with MutualImpurityReductionVariableSelection().

RFSurrogates 0.3.4

• var.select.smd(), var.select.md(), var.relations(), var.relations.mfi(): Made several improvements to developer experience:
  • create.forest now defaults to is.null(forest), so it will automatically be TRUE if no forest is provided, and FALSE otherwise.
  • x is no longer required if create.forest is FALSE.
  • (Internal) Inverted some nested guard clauses for readability.
• addLayer(): Refactor for-loop to lapply.
  • Add num.threads param to enable parallelization using parallel::mclapply(). It defaults to 1 for backward compatability.
• getTreeranger(): Refactor lapply() to parallel::mclapply().
  • Add num.threads param (passed to mc.cores in parallel::mclapply()). It defaults to 1 for backward compatability.
  • Add add_layer param to include the effect of addLayer within the same loop. Defaults to FALSE for backward compatability.
  • (Internal) getsingletree(): Add add_layer param to enable adding layers within the same loop.
• addSurrogates():
  • Clarified default value for num.threads to be parallel::detectCores() by adding it as a default to the parameter
  • Added assertion that RF is a ranger object.
  • Added assertion that RF$num.trees and length(trees) are equal. This is not considered a breaking change since these values should always be equal when the function is used correctly.
• Added S3 classes to the trees list objects.
  • getTreeranger() now returns a RangerTrees list.
  • addLayer() and getTreeranger(add_layer = TRUE) add the LayerTrees class to the list (indicating presence of the layer list item). It now requires that its trees param inherits RangerTrees.
  • addSurrogates() now adds the SurrogateTrees class. It now requires that its trees param inherits RangerTrees.

RFSurrogates 0.3.3

• Fixed meanAdjAgree() bug which caused mean adjusted agreement pairings to be set to NA incorrectly when variables was a subset or differently ordered than candidates.

RFSurrogates 0.3.2

• Moved to new repository: AGSeifert/RFSurrogates
• Fixed mean.index() bug which caused the return value to be of incorrect length in some cases (#4).

---

This is the legacy changelog of SurrogateMinimalDepth, previously located at Version_info.txt, adapted to Tidyverse Style for news. Future releases should begin following Semantic Versioning, and changes should be easier to track using Conventional Commits.

---

RFSurrogates 0.3.1 (RFSurrogates 0.3.1)

• Renamed package SurrogateMinimalDepth to RFSurrogates
• Renamed parameter ntree to num.trees throughout the package.

RFSurrogates 0.3.0 (SurrogateMinimalDepth 0.3.0)

• Round Mean Adjusted Agreement to 2 digits.
• meanAdjAgree(): Sped up.
• Changed that relations to variables not used as primary split are set as NA (before it was 0).
• var.relations.corr() renamed to var.relations.mfi() (as named in the paper).
• Bugfix for MFI and MIR.
• Permutation approach to determine p-values was optimized for the selection of important and related variables.
• Default approach for p-value calculation was set to "permutation".

RFSurrogates 0.2.1 (SurrogateMinimalDepth 0.2.1)

• Included possibility to analyze categorial variables.

RFSurrogates 0.2.0 (SurrogateMinimalDepth 0.2.0)

• New var.select.mir() to select important variables based on Mutual Impurity Reduction.
• New var.relations.corr() to calculated unbiased relations.
• Added multicore calculation of variable relations.
• var.relations(): Included possibility to only calculate relations and not select variables.
• Included the possibility to set case weights.

RFSurrogates 0.1.10 (SurrogateMinimalDepth 0.1.10)

• Fixed bug not including first and last surrogate.
• Added random selection of surrogates when adjusted agreement are the same.
• Fixed bug in var.relation() example.
• New build.clusters() to obtain variable groups.

RFSurrogates 0.1.9 (SurrogateMinimalDepth 0.1.9)

• var.select.smd(): Some specifics to set s were adapted.
• var.select.smd(), var.select.md(): Added save.memory parameter (to build the forst with ranger).

RFSurrogates 0.1.8 (SurrogateMinimalDepth 0.1.8)

• var.select.md() now executes var.select.smd() with s = 0 instead of using a separate function.
• New reduce.surrogates() function added.
• var.select.smd(): Added parameters create.forest and forest to use existing forests (e.g. created by reduce.surrogates()).
• Renamed parameter trees to forest containing trees and variable names.
• The C-code is updated to enable multicore analysis.

RFSurrogates 0.1.7 (SurrogateMinimalDepth 0.1.7)

• var.select.smd(), var.select.md(): Added parameter. save.ranger to save ranger object.

RFSurrogates 0.1.6 (SurrogateMinimalDepth 0.1.6)

• var.select.smd(): Adapted threshold for low depth trees.

RFSurrogates 0.1.5 (SurrogateMinimalDepth 0.1.5)

• var.select.smd(): Added s as a parameter.
• var.select.smd(), var.select.md(): Implemented survival function.

RFSurrogates 0.1.4 (SurrogateMinimalDepth 0.1.4)

• All errors and comments from coauthors implemented: first version uploaded on github.

RFSurrogates 0.1.3 (SurrogateMinimalDepth 0.1.3)

No changelog available.

RFSurrogates 0.1.2 (SurrogateMinimalDepth 0.1.2)

No changelog available.

RFSurrogates 0.1.1 (SurrogateMinimalDepth 0.1.1)

No changelog available.

RFSurrogates 0.1.0 (SurrogateMinimalDepth 0.1.0)

No changelog available.