R code for Enterprise mIner

[statman123]

Hi

I would like to get R syntax to put in the open source code in EM for the following models to compare

Decision Tree

Reg

Neural Net

Also the packages and libraries needed

thanks!

[MelodieRush]

Here is basic code for using the open source integration node for a decision tree in R. I use this code all the time 🙂

library(rpart)
&EMR_MODEL <- rpart(
&EMR_CLASS_TARGET ~ &EMR_CLASS_INPUT + &EMR_NUM_INPUT, data=&EMR_IMPORT_DATA,method ="class")

[MelodieRush]

Here is basic code for doing a logistic regression in R using the open source node in EM.

&EMR_MODEL <- glm(&EMR_CLASS_TARGET ~ &EMR_CLASS_INPUT + &EMR_NUM_INPUT,
                                     family=binomial(), data=&EMR_IMPORT_DATA)

[MelodieRush]

Basic code for Random Forest in R using the Open Source Node in EM.

#load library
library(randomForest)

#train model
&EMR_MODEL <- randomForest(&EMR_CLASS_TARGET ~ &EMR_CLASS_INPUT + &EMR_NUM_INPUT, ntree=100, mtry=5, data=&EMR_IMPORT_DATA, importance=TRUE)

[MelodieRush]

Basic code for neural networks in R using the Open Source node in EM.

library(nnet)
&EMR_MODEL <- nnet(&EMR_CLASS_TARGET ~ &EMR_CLASS_INPUT + &EMR_NUM_INPUT, size=2, rang=0.1, data=&EMR_IMPORT_DATA)

[MelodieRush]

Keep in mind the recommended version of R.
(taken from EM help)

R Configuration

Typically, R is installed on the same machine as the SAS Enterprise Miner Server. The Open Source Integration node is verified to work with 64-bit R 3.12, the pmml 1.4.2 package, and the XML 3.98-1.1 package. It is not recommended that you use 32-Bit R because SAS Enterprise Miner data sources often require large allocations of memory.