A full RStudio Server setup to use H2O in 5 minutes

RStudio Server
Go from zero to a full setup to fit models in a few clicks.

Mauricio “Pachá” Vargas S.


October 5, 2019

Updated 2022-05-28: Updated DO link.

At the end of the post there is a promotional link for free DigitalOcean credits.


At LatinR 2019 I had a nice conversation with Dr. LeDell that left me motivated to use H2O, so I updated my RStudio Server image on Digital Ocean to include H2O and ShinyML amongst other ML packages. Please find the update at https://marketplace.digitalocean.com/apps/rstudio.

The magic of that image is that you make 3 clicks to pick the hardware (i.e. 128 GB in RAM or 256) and in no more than 2-5 minutes you’ll have a server ready to rock. Compiling software is sometimes slow, I’ve done that for you.

With the server running you’ll only need to clone your repositories, train the models, and download the output. All of that from the browser in a pay just the time you use mode.


Here’s a video where I go from zero to variable importance in H2O. It includes creating the server, ssh login, creating a user, opening RStudio Server, cloning a repository and then fitting a model in just 06:08. No install.packages() required

Get free DigitalOcean credits

Please help me covering the hosting costs of Open Trade Statistics. If you register with this link, you’ll get free credits to try DigitalOcean and I also get credits for my trade project:

My DigitalOcean Referral Link

Buy me a coffee

If you like my work, buy me a coffee, it shall be used to produce more of it. Thank you for your support!