CORElearn (version 1.57.3)

predict.CoreModel: Prediction using constructed model

Description

Using a previously built model and new data, predicts the class value and probabilities for classification problem and function value for regression problem.

Usage

# S3 method for CoreModel
predict(object, newdata, ..., costMatrix=NULL, 
                            type=c("both","class","probability"))

Value

For regression model a vector of predicted values for given input instances. For classification problem the parameter type controls what is returned. With default value "both"

function returns a list with two components class

and probabilities containing predicted class values and probabilities for all class values, respectively. With type set to "class" or "probability" the function returns only the selected component as vector or matrix.

Arguments

object

The model structure as returned by CoreModel.

newdata

Data frame with fresh data.

costMatrix

Optional cost matrix can provide nonuniform costs for classification problems.

type

Controls what will be return value in case of classification.

...

Other model dependent options for prediction. See helpCore.

Author

Marko Robnik-Sikonja, Petr Savicky

Details

The function uses the object structure as returned by CoreModel and applies it on the data frame newdata. The newdata must be transformable using the formula specified for building the model (with dependent variable removed). If the dependent variable is present in newdata, it is ignored.

Optional cost matrix can provide nonuniform costs for classification problems. For regression problem this parameter is ignored. The costs can be different from the ones used for building the model in CoreModel.

See Also

CORElearn, CoreModel, modelEval, helpCore, paramCoreIO.

Examples

# use iris data set

# build random forests model with certain parameters
modelRF <- CoreModel(Species ~ ., iris, model="rf", 
              selectionEstimator="MDL",minNodeWeightRF=5,rfNoTrees=100)
print(modelRF)

# prediction with node distribution
pred <- predict(modelRF, iris, rfPredictClass=FALSE, type="both")
# print(pred)

destroyModels(modelRF) # clean up