We are pleased to announce the general availability of Oracle Machine Learning for R 2.0 on Oracle Database 19c and 21c. Last year, we announced OML4R 2.0 for Oracle Autonomous Database. With this release, users can now connect to Oracle Database, Oracle Database Cloud Service, and Oracle Autonomous Database using the same standalone OML4R client package. OML4R 2.0 expands the set of native in-database algorithms exposed through R to include Neural Network, Random Forest, Exponential Smoothing Models, and XGBoost. Note, XGBoost is available from 21c and later.
Oracle Machine Learning for R (OML4R) leverages the database as a high-performance computing environment to explore, transform, and analyze data faster and at scale, while allowing the use of familiar R syntax and semantics. The in-database parallelized machine learning algorithms are exposed through a well-integrated R interface.
Data scientists and other R users can also store and manage user-defined R functions as well as R objects directly in the database – as opposed to being managed in flat files. These features facilitate collaboration across the data science team by enabling convenient hand-off of data science work products from data scientists to application developers for immediate deployment. This release of OML4R uses Oracle R Distribution 4.0.5, based on R 4.0.5.
Get started today!
As always, you can use OML4R 2.0 on Oracle Autonomous Database through Oracle Machine Learning Notebooks. In your Autonomous Database instance, create a notebook, specify the R interpreter in a paragraph using ‘%r’, load the ORE library, and you’re ready to go. See the template examples available through OML Notebooks on Autonomous Database for over 35 notebooks using OML4R functionality for data preparation and exploration, modeling, and deployment.
For on-premises Oracle Database and Database Cloud Service users, as well as to use Autonomous Database from third-party IDEs, install the OML4R client. To install OML4R 2.0 for use with your Oracle Database or Database Cloud Service instances, install the OML4R database server components.
Highlighted enhancements in OML4R 2.0
In addition to the in-database algorithms that are now part of the OML4R API (Neural Network, Random Forest, Exponential Smoothing Models, and XGBoost), here are a few additional enhancements in OML4R 2.0.
The in-database algorithms support prediction details to understand which predictors most contribute to an individual prediction. The predict method allows users to request the top N attributes using argument topn_attrs to return those predictors, which contains the predictor name, value, and an associated weight. Also enabled for prediction is to return the top N predicted classes for classification models.
The datastore functionality, which allows user to store and retrieve R and OML4R objects directly in the database, enables easily renaming existing datastores and their contained objects. You can also remove datastore entries in batch using convenient selection through name pattern matching.
Similarly, in the R script repository for storing and managing user-defined R functions that can also be used with OML4R Embedded R Execution, users can load and drop scripts using selection through name pattern matching.
Availability
The OML4R 2.0 standalone client for use with Oracle Database and Oracle Autonomous Database, and the server components for use with Oracle Database are available for download here.
Features and benefits
- Use Oracle Database and Oracle Autonomous Database as a high-performance computing environment
- Connect to Oracle Database and Oracle Autonomous Database from third-party IDEs using a common standalone OML4R client package
- Explore, transform, and analyze data faster and at scale
- Eliminate or minimize data movement
- Use in-database parallelized and distributed machine learning algorithms
- Build more models on more data, and score large volume data faster
- Use in-database algorithms via a convenient R API
- Increase productivity using in-database algorithm features including automatic data preparation, partitioned models, and integrated text mining
- Run user-defined R functions in database spawned and managed R engines
- Store and manage R objects and user-defined R functions in the database
- Collaborate: easily hand-off data science work products from data scientists to developers
- Run user-defined functions using system-supported data-parallelism and task-parallelism
- Return structured and image results in R, SQL, and (on Autonomous Database) REST APIs
Support for the R community
Oracle first introduced support for R on Oracle Database in 2011 and contributed to and maintains the ROracle package (on CRAN) for Oracle Database connectivity. Oracle is a founding member and contributor to the R Consortium. The central mission of the R Consortium is to work with and provide support to the R Foundation and to the key organizations developing, maintaining, distributing and using R software through the identification, development and implementation of infrastructure projects.