After this first lab every student should have installed and working on their laptops:
- R
- RStudio
- some R packages TBD
- Java (needed for h2o)
- h2o (from R, run this to test if it is working)
should have
- a github account
- an AWS account with $100 worth of free student credits
should be able to
- create a simple Rmarkdown file, run it in RStudio and get html report
- upload files to their github account
- start an EC2 instance, create an image, stop/restart, configure security groups (open ports)
- use
ssh
and basic Unix/Linux command line - do same work in RStudio server on EC2 and in RStudio locally on laptop
On EC2 one can use this AMI with R and RStudio server already
installed. Please change the default password for the rstudio
user (e.g. as described
at the previous link).
Warning: The Rstudio connection (from your browser) does not use SSL, therefore don't use this server on projects requiring security (e.g. with confidential data) or set up SSL connection (https) before you connect to RStudio for the very first time.
Install Java (required by h2o) via command line (e.g. by logging in via ssh):
sudo apt-get install default-jre
(then you can install h2o as an R package from RStudio).
install.packages("devtools")
devtools::install_github("Microsoft/LightGBM", subdir = "R-package")
(might need cmake: sudo apt-get install cmake
)
AMI with CUDA, cuDNN. Install RStudio keras. Other DL libs?