Title: | Fit and Deploy DECORATE Trees |
---|---|
Description: | DECORATE (Diverse Ensemble Creation by Oppositional Relabeling of Artificial Training Examples) builds an ensemble of J48 trees by recursively adding artificial samples of the training data ("Melville, P., & Mooney, R. J. (2005) <DOI:10.1016/j.inffus.2004.04.001>"). |
Authors: | Matthias Bogaert |
Maintainer: | Matthias Bogaert <[email protected]> |
License: | GPL (>=2) |
Version: | 0.1.2 |
Built: | 2024-09-25 03:14:06 UTC |
Source: | https://github.com/matthbogaert/decorater |
DECORATE
(Diverse Ensemble Creation by Oppositional Relabeling of
Artificial Training Examples) builds an ensemble of J48 trees by recursively
adding artificial samples of the training data.
DECORATE(x, y, C = 15, I = 50, R = 1, verbose = FALSE)
DECORATE(x, y, C = 15, I = 50, R = 1, verbose = FALSE)
x |
a data frame of predictor (numeric, integer or factors). Character variables should be transformed to factors. |
y |
a vector of response labels. Only {0, 1} is allowed. |
C |
the desired ensemble size. Set to 15 as recommended by Melville and Mooney (2005). |
I |
the maximum number of iterations. Set to 50 as recommended by Melville and Mooney (2005). |
R |
the amount of articially generated examples, expressed as a fraction of the number of training examples. R is set to 1, meaning that the number of artificially created samples is equal to the training set size. |
verbose |
TRUE or FALSE. Should information be printed on the screen? |
an object of class DECORATE
.
Authors: Matthias Bogaert, Maintainer: [email protected]
Melville, P., & Mooney, R. J. (2005). Creating diversity in ensembles using artificial data. Information Fusion, 6(1), 99-111. <doi: 10.1016/j.inffus.2004.04.001>
data(iris) y <- as.factor(ifelse(iris$Species[1:100]=="setosa",0,1)) x <- iris[1:100,-5] dec <- DECORATE(x = x, y = y)
data(iris) y <- as.factor(ifelse(iris$Species[1:100]=="setosa",0,1)) x <- iris[1:100,-5] dec <- DECORATE(x = x, y = y)
Prediction of new data using DECORATE
## S3 method for class 'DECORATE' predict(object, newdata, type = "prob", all = FALSE, ...)
## S3 method for class 'DECORATE' predict(object, newdata, type = "prob", all = FALSE, ...)
object |
an object of the class |
newdata |
a data frame containing the same predictors as in the training phase. |
type |
character specifying whether to return the probabilites ('prob') or class ('class'). Default: prob. |
all |
Return the predictions per tree instead of the average (default = FALSE). |
... |
Not used currently. |
vector containing the response probabilities.
data(iris) y <- as.factor(ifelse(iris$Species[1:100]=="setosa",0,1)) x <- iris[1:100,-5] dec <- DECORATE(x = x, y = y) predict(object=dec,newdata=x)
data(iris) y <- as.factor(ifelse(iris$Species[1:100]=="setosa",0,1)) x <- iris[1:100,-5] dec <- DECORATE(x = x, y = y) predict(object=dec,newdata=x)