Data Science Package for R v7.5

Suggest edits

The Data Science Package for R is an EDB-provided package that bundles a curated collection of R data science, statistics, and machine learning packages for use with the WarehousePG PL/R (PL/R) procedural language.

The package compiles from source against the R runtime bundled by the WarehousePG PL/R Language Extension and requires PL/R to be installed and enabled on your cluster.

Available packages

The Data Science Package for R bundles 162 curated R packages, including the transitive dependencies required to build and load them.

CategoryPackages
ML and modelingcaret, randomForest, e1071, glmnet, adabag, ipred, gbutils, neuralnet, recipes, ModelMetrics, pROC, ROCR
Statisticslme4, MASS, car, quantreg, sandwich, lmtest, survival, mvtnorm, coda, MCMCpack, mcmc, arm, fBasics, fGarch
Time seriesforecast, tseries, zoo, xts, TTR, quantmod, urca, fracdiff, timeSeries, timeDate, MTS
Data wranglingdplyr, tidyr, data.table, plyr, reshape2, purrr, broom, readr, vroom, lubridate, stringr, stringi
Visualizationggplot2, scales, RColorBrewer, gplots, cowplot, diagram, dichromat, viridisLite
Graphs and clusteringigraph, flashClust, compHclust, hybridHclust, fastICA, proxy
Infrastructure and databaseDBI, RPostgreSQL, Rcpp, RcppArmadillo, RcppEigen, foreach, doParallel, future, XML, jsonlite, curl

The Data Science Package for R pins every package to a snapshot of the Comprehensive R Archive Network (CRAN) dated 2026-07-01, for reproducibility. Three packages, compHclust, hybridHclust, and plogr, are no longer available as current releases on CRAN, so they're pulled from the CRAN Archive instead, at the last version each package ever published.

For the exact version of any package in the bundle, check the manifest that ships inside the package at $GPHOME/ext/DataScienceR/MANIFEST, with one package==version entry per line. Comparing the manifest between two versions of the Data Science Package for R shows you exactly what changed.

Prerequisites

Before installing the Data Science Package for R, ensure that:

  • The edb-whpg7-plr package is installed and PL/R is enabled on your cluster. See WarehousePG PL/R Language Extension.
  • You have sourced /usr/edb/whpg7/greenplum_path.sh and the $COORDINATOR_DATA_DIRECTORY and $GPHOME environment variables are set.
Note

Because the WarehousePG PL/R Language Extension doesn't currently include cairo support, plotting functions that render through a graphics device, for example ggplot2 output saved with png(), aren't supported.

Downloading and installing the Data Science Package for R

The Data Science Package for R is available for RHEL 8 and RHEL 9. Install it on each host in your WarehousePG cluster.

  1. From the coordinator, download the package from the EDB repository:

    export EDB_SUBSCRIPTION_TOKEN=<your-token>
export EDB_REPO=gpsupp
curl -1sSLf "https://downloads.enterprisedb.com/$EDB_SUBSCRIPTION_TOKEN/$EDB_REPO/setup.rpm.sh" | sudo -E bash
sudo dnf download edb-whpg7-data-science-r

    Where <your-token> is your EDB subscription token.

  2. Create a file all_hosts on the coordinator that lists all hosts in the cluster:

    cdw
scdw
sdw1
sdw2
sdw3

  3. Use gpsync to transfer the package to all hosts, then use gpssh to install it:

    gpsync -f all_hosts <package-name> =:/tmp
gpssh -f all_hosts -e 'sudo dnf install -y /tmp/<package-name>'

    Where <package-name> is the name of the package file you downloaded.

No cluster restart is needed after installing the package.

After installation, the packages are available at $GPHOME/ext/DataScienceR/library. The package also installs a pinned manifest at $GPHOME/ext/DataScienceR/MANIFEST, with one package==version entry per line, and a $GPHOME/ext/DataScienceR/VERSION file that records the package's own version.

Overriding individual package versions

The Data Science Package for R lets you override the version of any single bundled package after installation, without modifying any files owned by the package. To check which version of a package is currently bundled before overriding it, see Available packages.

The bundled R resolves packages through a two-tier R_LIBS_SITE search path, set by the package's Renviron.site file:

R_LIBS_SITE=$GPHOME/ext/DataScienceR/overrides:$GPHOME/ext/DataScienceR/library
  • library holds the package-owned bundle of 162 packages. Don't modify it directly.
  • overrides is empty by default and belongs to you. Any package version you install there takes precedence over the bundled version.

To pin a different version of a package, for example rolling abind back to an older release:

  1. On the coordinator, make the bundled R's helper scripts executable. Making them executable is a one-time step, needed because current edb-whpg7-plr packages ship them without execute permissions:

    sudo chmod +x $GPHOME/lib64/R/bin/*

  2. Download the source tarball for the version you want, from CRAN or, for a version no longer current, from the CRAN Archive:

    curl -LO https://cran.r-project.org/src/contrib/Archive/abind/abind_1.4-5.tar.gz

  3. Build and install the desired version into the overrides directory, using the bundled R:

    $GPHOME/lib64/R/bin/R CMD INSTALL --library=$GPHOME/ext/DataScienceR/overrides abind_1.4-5.tar.gz

    The override must be compiled against the bundled R. Build it once on the coordinator, then copy the result everywhere. Don't rebuild it separately on each host.

  4. Copy the overrides directory to every host in the cluster, for example with gpsync:

    gpsync -f all_hosts $GPHOME/ext/DataScienceR/overrides =:$GPHOME/ext/DataScienceR/

New sessions on the coordinator and on all segments load the pinned version, and no cluster restart is needed. After installing an override, recycle any connection pools, since sessions that are already connected keep the package version they loaded until they reconnect.

To revert to the bundled version, remove the package's directory from overrides on every host:

gpssh -f all_hosts -e "rm -rf $GPHOME/ext/DataScienceR/overrides/abind"

New sessions fall back to the bundled version. rpm -V stays clean, and upgrading the package doesn't delete your overrides.

Uninstalling the Data Science Package for R

Uninstall the package on all hosts:

gpssh -f all_hosts -e 'sudo dnf remove -y edb-whpg7-data-science-r'

No cluster restart is needed to complete the removal.

Note

After uninstalling the Data Science Package for R, any user-defined functions (UDFs) that you've created that use packages from this bundle return an error.

← Prev

Data Science Package for Python

↑ Up

Installing Optional Extensions

Next →

Installing Extensions