bigcforest-class: Classification Random Forests
Description
Class representing a classification random forest.
Objects from the Class
Objects can be created by calls of the form new("bigcforest", ...), but most often are generated by bigrfc.Slots
.Data:- Object of class
"list". Each element is a "bigctree", representing a tree in the random classification forest. nexamples:- Object of class
"integer". Number of examples in the training set (including synthesized examples for unsupervised learning). varselect:- Object of class
"integer". Indices of the columns of x that were used to train the model. factorvars:- Object of class
"logical". Indicates which variables are factors or categorical (TRUE)), and which are numeric (FALSE). varnlevels:- Object of class
"integer". Number of levels in each categorical variable, or 0 for numeric variables. contvarseq:- Object of class
"integer". Maps the continuous variables in varselect to the columns in big.matrix a. Meant for internal use by bigrfc or grow when growing trees. y:- Object of class
"factor". Class labels for the training set. ytable:- Object of class
"table". Counts of training examples in each class. yclasswts:- Object of class
"matrix". One-dimensional matrix of scaled weights for each class. ntrees:- Object of class
"integer". Number of trees in the forest. nsplitvar:- Object of class
"integer". Number of variables to split on at each node. maxndsize:- Object of class
"integer". Maximum number of examples in each node when growing the trees. maxeslevels:- Object of class
"integer". Maximum number of levels for categorical variables for which exhaustive search of possible splits will be performed. nrandsplit:- Object of class
"integer" Number of random splits to examine for categorical variables with more than maxeslevels levels. oobtimes:- Object of class
"integer". Number of times each training example has been out-of-bag. oobvotes:- Object of class
"matrix". Out-of-bag votes for each training example. oobpred:- Object of class
"integer". Out-of-bag predictions for each training example. trainclserr:- Object of class
"numeric". Training errors of out-of-bag examples, by class. trainerr:- Object of class
"numeric". Total training error of out-of-bag examples. trainconfusion:- Object of class
"table". Confusion matrix for out-of-bag examples. varginidec:- Object of class
"numeric". Decrease in Gini impurity for each variable over all trees. cachepath:- Object of class
"character.or.NULL". Path to folder where data caches used in building the forest were stored, or NULL if data was processed completely in memory.
Extends
Class "list", from data part.
Class "vector", by class "list", distance 2.Methods
- grow
signature(forest = "bigcforest"): Grow more trees in the random forest, using the same parameters. See grow for details.- merge
signature(x = "bigcforest", y = "bigcforest"): Merge two random forests into one. See merge for details.- predict
signature(object = "bigcforest"): Predict the classes of a set of test examples. See predict for details.- varimp
signature(forest = "bigcforest"): Compute variable importance based on out-of-bag estimates. See varimp for details.- fastimp
signature(forest = "bigcforest"): Compute fast (Gini) variable importance. See fastimp for details.- interactions
signature(forest = "bigcforest"): Compute variable interactions. See interactions for details.- proximities
signature(forest = "bigcforest"): Compute the proximity matrix. See proximities for details.- prototypes
signature(forest = "bigcforest", prox = "bigrfprox"): Compute class prototypes. See prototypes for details.- show
signature(object = "bigcforest"): Print the random forest.- summary
signature(object = "bigcforest"): Print summary information on the random forest, including out-of-bag training error estimates and the confusion matrix.