This collection of samples will get you started on R, Microsoft R Server and R Tools for Visual Studio. To get them:
README files will help you navigate the samples.
At the top level, A First Look at R gives a gentle introduction for newcomers to R. MRS and Machine Learning gives examples of how to use R and Microsoft R Server for machine learning.
Reproducible R Toolkit, which ensures that the libraries you used to build your R program are always available to others that want to reproduce your work.
Microsoft R Server is an extension of R that allows you to handle more data and handle it faster. It gives R two powerful capabilities:
Larger data sets. MRS can process out-of-memory data from a variety of sources including Hadoop clusters, databases and data warehouses. You never have to be limited by your RAM again.
Parallel, multi-core processing. MRS can efficiently distribute computation across all the computational resources it has available. On your personal workstation or a remote cluster, MRS will get an answer faster.
Figure 1. MRS and MRO with MKL have significantly better computation performance related to certain matrix calculation than R and MRO without MKL. Simulated data is used in this calculation. For a technical comparison of R with MRO and MRS, check out Lixun Zhang’s detailed discussion on the topic.
Figure 2. This figure compares elapsed time in seconds used in building Logistic Regression models to predict whether the arrival of scheduled passenger flights will be delayed by more than 15 minutes. Elapsed time used in CRAN R increases dramatically when increasing a small number of rows, while MRS only increases by approximately 2 times. For details of this benchmark, check out
R MRO MRS Comparison This six-part comparison shows where the commands, syntax, constructs and performance of R, Microsoft R Open and Microsoft R Server are similar, and where they differ.
Machine learning Samples learning to predict flight delays, housing prices and bike rentals show how to solve real world problems with both R and MRS. They also show you how to use several popular machine learning models and how to deploy them as an Azure Web Service using an Azure Machine Learning workspace.
Benchmarks Microsoft R Open includes the Intel Math Kernel Library (MKL) for fast, parallel linear algebra computations. This example runs a number of compute-intensive benchmarks to show the performance gains that are possible through the use of MKL.
Figure 3. With simulated data, using 2 threads tends to give better performance than using 1 thread for certain matrix related calculations. Check out
MRO-MKL-benchmarks.R example for more details.