dvr_ensemble: Dependent Variable Regression Ensemble

In dannymorris/outsiders: Multidimensional Outlier Detection

Description

Dependent Variable Regression Ensemble

Usage

dvr_ensemble(formula, data, method = "lm", n_predictions, n_train_points,
  score_set = c("all", "test"), error_agg_fun = mean, scores_only = TRUE,
  ...)

Arguments

formula	a formula interface specifying the model (see help("lm") for more detail)
data	a matrix or data.frame containing variables in model
method	a model function (e.g. "lm", "randomForest")
n_predictions	an integer specifying the number of components in the ensemble. If score_set = "test", set this high enough to ensure all points are predicted a sufficient number of times
n_train_points	an integer or numeric value specifying the number of rows used in the training phase of each ensemble
score_set	one of "all" or "test". If "all", scores all N points in each training iteration. If "test", score out of sample points in each iteration.
error_agg_fun	a function for combining the squared prediction errors. Defaults to mean.
scores_only	logical, if TRUE return a vector of outlier scores. If FALSE, return the error matrix and outlier scores

Value

if scores_only = TRUE, a vector of outlier scores. If FALSE, a list with outlier scores and the ensemble error matrix

References

section 3.2.1 of "Outlier Analysis" (C. C. Aggarwal. Outlier Analyis. Springer, 2017.)

Examples

ensemble_lm(formula = Sepal.Length ~ ., data = iris[,-5],
n_predictions=100, n_train_points=100, error_agg_fun = median, 
scores_only = T)

dannymorris/outsiders documentation built on May 13, 2019, 1:22 p.m.