mlmeta

Machine Learning Metaprogramming for R

These functions export several kinds of machine learning models trained in R to other programming languages, so the models can score data in a separate environment without R. Currently Base SAS is supported, and additional SAS licenses for SAS/IML, SAS/STAT, and SAS Enterprise Miner are not required. In some cases the models in SAS run faster and use less memory than in R.

Supported models:

• Bagged MARS from caret
• Conditional inference trees from party
• Random forests from party
• Multivariate Adaptive Regression Splines (MARS) from earth
• Generalized Boosted Regression Models from gbm
• Artificial neural networks from nnet

Later I will post this package on CRAN. Until then, use:

require(devtools)
install_github("az0/mlmeta")

For an example usage, please see:

• Train neural network in R, predict in SAS
• Model decision tree in R, score in Base SAS

If you find these useful, please consider a donation to Compassion International. Even a penny will help.