Setup for data science related dev on (>=) RHEL 7.6 (Maipo) and CentOS

data-science-setup.sh

# Need to be sudo to install the following
yum update
yum install -y wget bzip2 tmux

# Download and run anaconda3 for python3 and follow instructions
wget https://repo.continuum.io/archive/Anaconda3-5.3.1-Linux-x86_64.sh | sh

# Conda install essential datascience packages
conda update --all
conda install pandas numpy scipy scikit-learn keras tensorflow pytorch nltk matplotlib seaborn

# Source for installing R
# https://bluehatrecord.wordpress.com/2014/10/13/installing-r-on-red-hat-enterprise-linux-6-5/

# To install R, under [rhui-REGION-rhel-server-optional]
# change enabled=0 to enabled=1 in the following file
vi /etc/yum.repos.d/redhat-rhui.repo
yum install R

# Might have to additionally install the following
# yum install libssh2-devel libgit2-devel libxml2-devel

# Make sure the paths of $locate libicuio.so (generally stored in ~/anaconda3/lib/)
# is in PATH of root user, else export that directory to PATH
# export PATH=$PATH:~/anaconda3/lib/

# Fixes issues like `error while loading shared libraries: libicui18n.so.58: cannot open shared object file: No such file or directory`
# Fixes issues like `package ‘xml2’ had non-zero exit status`

# Install essential datascience packages in R
R -e 'install.packages(c("tidyverse", "devtools", "ggplot2", "forecast", "IRkernel", "lubridate", "caret", "doSNOW", "foreach", "doParallel", "Metrics"))'

# Command to make R kernel available in Jupyter Notebook
R -e 'IRkernel::installspec(name="ir3.5", displayname="R35"); devtools::install_github("ellisp/forecastxgb-r-package/pkg")'