RevoScaleR Functions for Spark on Hadoop

The RevoScaleR package provides a set of portable, scalable, distributable data analysis functions. This page presents a curated list of functions that might be particularly interesting to Hadoop users. These functions can be called directly from the command line.

The RevoScaleR package supports two Hadoop compute contexts:

• RxSpark (recommended), a distributed compute context in which computations are parallelized and distributed across the nodes of a Hadoop cluster via Apache Spark. This provides up to a 7x performance boost compared to RxHadoopMR. For guidance, see How to use RevoScaleR on Spark.

• RxHadoopMR (deprecated), a distributed compute context on a Hadoop cluster. This compute context can be used on a node (including an edge node) of a Cloudera or Hortonworks cluster with a RHEL operating system, or a client with an SSH connection to such a cluster. For guidance, see How to use RevoScaleR on Hadoop MapReduce.

On Hadoop Distributed File System (HDFS), the XDF file format stores data in a composite set of files rather than a single file.

Data Analysis Functions

Import and Export Functions

Function Name		Description	Help
rxDataStep		Transform and subset data. Creates an .xdf file, a comma-delimited text file, or data frame in memory (assuming you have sufficient memory to hold the output data) from an .xdf file or a data frame.	<small>View</small>
RxXdfData		Creates an efficient XDF data source object.	<small>View</small>
RxTextData		Creates a comma delimited text data source object.	<small>View</small>
rxGetInfo		Retrieves summary information from a data source or data frame.	<small>View</small>
rxGetVarInfo		Retrieves variable information from a data source or data frame.	<small>View</small>
rxGetVarNames		Retrieves variable names from a data source or data frame.	<small>View</small>
rxHdfsFileSystem		Creates an HDFS file system object.	<small>View</small>

Function Name		Description	Help
rxDataStep		Transform and subset data. Creates an .xdf file, a comma-delimited text file, or data frame in memory (assuming you have sufficient memory to hold the output) from an .xdf file or a data frame.	<small>View</small>
rxFactors		Create or recode factor variables in a composite XDF file in HDFS. A new file must be written out.	<small>View</small>

Function Name		Description	Help
rxQuantile		Computes approximate quantiles for .xdf files and data frames without sorting.	<small>View</small>
rxSummary		Basic summary statistics of data, including computations by group. Writing by group computations to .xdf file not supported.	<small>View</small>
rxCrossTabs		Formula-based cross-tabulation of data.	<small>View</small>
rxCube		Alternative formula-based cross-tabulation designed for efficient representation returning ‘cube’ results. Writing output to .xdf file not supported.	<small>View</small>

Function Name		Description	Help
rxLinMod		Fits a linear model to data.	<small>View</small>
rxLogit		Fits a logistic regression model to data.	<small>View</small>
rxGlm		Fits a generalized linear model to data.	<small>View</small>
rxCovCor		Calculate the covariance, correlation, or sum of squares / cross-product matrix for a set of variables.	<small>View</small>
rxDTree		Fits a classification or regression tree to data.	<small>View</small>
rxBTrees		Fits a classification or regression decision forest to data using a stochastic gradient boosting algorithm.	<small>View</small>
rxDForest		Fits a classification or regression decision forest to data.	<small>View</small>
rxPredict		Calculates predictions for fitted models. Output must be an XDF data source.	<small>View</small>
rxKmeans		Performs k-means clustering.	<small>View</small>
rxNaiveBayes		Fit Naive Bayes Classifiers on an .xdf file or data frame for small or large data using parallel external memory algorithm.	<small>View</small>

Compute Context Functions

Function Name		Description	Help
RxHadoopMR		Creates an in-data, file-based Hadoop compute context.	<small>View</small>
RxSpark		Creates an in-data, file-based Spark compute context. Computations are parallelized and distributed across the nodes of a Hadoop cluster via Apache Spark.	<small>View</small>
rxSparkConnect		Creates a persistent Spark compute context.	<small>View</small>
rxSparkDisconnect		Disconnects a Spark session and return to a local compute context.	<small>View</small>
rxInstalledPackages		Returns the list of installed packages for a compute context.	<small>View</small>
rxFindPackage		Returns the path to one or more packages for a compute context.	<small>View</small>

Data Source Functions

Of course, not all data source types are available on all compute contexts. For the Hadoop compute contexts, two types of data sources can be used.

Function Name		Description	Help
RxXdfData		Creates an efficient XDF data source object.	<small>View</small>
RxTextData		Creates a comma delimited text data source object.	<small>View</small>
RxHiveData		Generates a Hive Data Source object.	<small>View</small>
RxParquetData		Generates a Parquet Data Source object.	<small>View</small>
rxSparkDataOps		Lists cached RxParquetData or RxHiveData data source objects.	<small>View</small>
rxSparkRemoveData		Removes cached RxParquetData or RxHiveData data source objects.	<small>View</small>

The Hadoop compute context has a number of helpful functions used for high performance computing and distributed computing. Learn more about the entire set of functions in the Distributed Computing guide.

RevoScaleR also provides some wrapper functions for accessing Hadoop/HDFS functionality via R. These functions require access to Hadoop, either locally or remotely via the RxHadoopMR or RxSpark compute contexts.