Saves/loads random forests model to/from file
saveRF: the internal structure of given random forests model is saved to file. loadRF: the internal structure of random forests model is loaded from given file and a model is created and returned.
saveRF(model, fileName) loadRF(fileName)
model: The model structure as returned by CoreModel.fileName: Name of the file to save/load the model to/from.The function saveRF saves the internal structure of given random forests model to file. The structures from C++ code are stored to the file with specified file, while internal structures from R are stored to file named fileName.Rda. The model must be a valid structure returned by CoreModel.
The function loadRF loads the internal structure of random forests saved in a specified files and returns access to it.
saveRF invisibly returns some debugging information, while loadRF
returns a loaded model as a list, similarly to CoreModel.
Marko Robnik-Sikonja
CORElearn, CoreModel.
# use iris data set # build random forests model with certain parameters modelRF <- CoreModel(Species ~ ., iris, model="rf", selectionEstimator="MDL",minNodeWeightRF=5, rfNoTrees=100, maxThreads=1) print(modelRF) # prediction with node distribution pred <- predict(modelRF, iris, rfPredictClass=FALSE, type="both") # print(pred) # saves the random forests model to file saveRF(modelRF, "tempRF.txt") # restore the model to another model loadedRF = loadRF("tempRF.txt") # prediction should be the same predLoaded <- predict(loadedRF, iris, rfPredictClass=FALSE, type="both") # print(predLoaded) # sum of differences should be zero subject to numeric imprecision sum(pred$probabilities - predLoaded$probabilities) cat("Are predicted classes of original and retrieved models equal? ", all(pred$class == predLoaded$class), "\n" ) # cat("Are predicted probabilities of original and retrieved model equal? ", # all(pred$probabilities == predLoaded$probabilities), "\n" ) # clean up the models when no longer needed destroyModels(modelRF) destroyModels(loadedRF) # clean up for the sake of R package checks file.remove("tempRF.txt") file.remove("tempRF.txt.Rda")