RevoScaleR is a package of machine learning functions, provided by Microsoft, that supports data science at scale.
Functions support data import, data transformation, summarization, visualization, and analysis.
At scale means that operations can be performed against very large datasets, in parallel, and on distributed file systems. Algorithms can operate over datasets that do not fit in memory, by using chunking and by reassembling results when operations are complete.
RevoScaleR provides many improvements over open source R functions. There are RevoScaleR functions corresponding to many of the most popular base R functions. RevoScaleR functions are denoted with an rx or Rx prefix to make them easy to identify.
RevoScaleR serves as a platform for distributed data science. For example, you can use the RevoScaleR compute contexts and transformations with the state-of-the-art algorithms in MicrosoftML. You can also use rxExec to run base R functions in parallel.
For examples of RevoScaleR in action, see these blogs:
How to get RevoScaleR
The RevoScaleR package for R is installed for free in Microsoft R Client. If you have Machine Learning Server or are using R in SQL Server, RevoScaleR is included by default.
If you are using Python, the revoscalepy package provides equivalent functionality.
The RevoScaleR package cannot be downloaded or used independently of the products and services that provide it.
Use RevoScaleR in SQL Server
These tutorials and samples demonstrate how to use RevoScaleR functions to get data from SQL Server, build models, and save models to a database for scoring.
- Learn to use compute contexts
- R for SQL developers: Train and operationalize a model
- Microsoft product samples on GitHub
Learn more about RevoScaleR
These tutorials demonstrate the use of RevoScaleR in other compute contexts supported by Machine Learning Server, including Hadoop.