The schedule for #HarvardDataFest is jam packed with awesome talks. I had to make a lot of tough choices today.

Unfortunately, I had to run out of Data Visualization with Tableau with Jess Cohen-Tanugi to take a call but she did a fabulous job showing how to visualize storms data (hurricanes, tropical storms, etc.), creating maps and “blending” data sources. I plopped myself down next to Alyssa Goodman and I'm so glad I did. She had downloaded the storms data and was playing with it in Glue. After the talk, she gave a 15 minute demo to me and Jess and I was amazed. I had seen Glue years ago (actually, I spun up the initial VMs when working at FAS Research Computing back in 2011 or so) and hadn't realized how much it had matured, how it's no longer a tool only for astrophysics. Glue is a visualization tool for all of science, and it's open source!

Next up was Reproducible & Collaborative Data Science with the Renku Platform with Rok Roskar. Having worked with Rok on getting data out of and into Dataverse I wasn't expecting to be surprised but the Renku first steps docs he lead us through talk about Dataverse in step 2! I was so excited I couldn't resist tweeting about it.

I stopped by Interactive Data Visualization with Shiny in R with Ista Zahn and the shiny_workshop materials he was leading us through looked fantastic but I don't have an immediate need for Shiny (and wasn't sure I wanted three hours of it) and wanted to see what else was cooking. I doubted Ista would be offended. He's a friend.

I caught the tail end of Natural Language Processing with Python with Elizabeth Piette and it seemed really interesting. Afterwards I chatted with a research computing consultant from Harvard Medical School (HMS) and gave him a demo of launching RStudio in Sid and he was suitably impressed. I also described the rsync-enabled installation of Dataverse that I expect will launch soon at HMS.

In Scaling Geospatial Analyses on Harvard's High Performance Computing Cannon Cluster Ben Lewis and Devika Kakkar have a really interesting talk on using Singularity on top of FAS RC's new Cannon cluster to get some serious geospatial computation done.

Tomorrow I'm looking forward to welcoming Fernando Pérez to Harvard.