Some Reflection on Workshop Version #1

[stephhazlitt]

General

Things that worked well

• software installation with support session in the AM seemed to work well re: getting started quickly
• dual screens throughout the room (great idea by @bevingtona to use Lync for this)
• progression of content was good from vector to raster to big data raster analysis
• I think the level of content overall was appropriate for the learners (but course evaluations will tell us for sure)
• Practical use of git/github in the hackathon was good

Things that could be improved

• optimizing font size for screens but still being able to use your laptop
• switching btw RStudio & slides without making people dizzy?
  • ideally could have separate screens - one with presentation, one with code
  • a webcast of your screen could also be useful

Day 1 — Git + GitHub + Project Management with RStudio

Things that worked well

• amount of content was about right for the afternoon time slot?

Things that could be improved

• for bcgov PC workstations, set home to C: Desktop before setting git config specs
• generating and editing text file with Terminal could be omitted (e.g. deviate from SWC module)
• more "your turn" modules — maybe more practice pushing/pulling to GitHub?
• some more discussion on GitHub as a project management tool. Issues, Pull requests, Milestones etc.

Day 2 — Vector & Raster Manipulation & Visualization

• In general, all modules had too much content. Likely that the material (including the stuff we didn't get to) would make a good two-day course, with a bit of extra work

Things that worked well

• pacing of content was 👌
• flow worked well: foundations of vector (sf, CRS, etc) -> practical vector -> foundations of raster -> practical raster

Things that could be improved

• Possibly spent too much time on CRS theory??
• using a new clean R code vs running prepped lines
  • if we do provide prepped lines, we should have a specially made R file for users to work off. Getting them to navigate a .Rmd file built for slides was friction-y.
• devote more time to both vector and raster handling (?)
• filepaths with Rmd (knit vs running interactive)
• avoid zipped files for data
• avoid teaching with .Rmd unless we explicitly introduce it as a concept.
• pick a common plot platform and apply it across lessons. In this case I was wishing we'd chosen {tmap} so that people and examples could draw across the whole lesson
• a brief introduction to RStudio paying particular attention to environments, relative paths and folder structure

Day 4 — Geospatial hackathon

Things that worked well

• High engagement, people seemed to enjoy it
• git lessons became solidified with a day of practical use
• regularly plugging the hackathon throughout the event

Things that could be improved

• Organization was a bit seat-of-pants - the switch from voting on projects to choosing projects was not smooth. Probably would leverage rOpensci example more and get all champions to stand around the room with a sign for their project (if they wanted to lead it) and have people go to the project they wanted. I think the process we used made the pool of projects too small.
• Could have used a couple more projects - teams slightly too large. This was made difficult because high interest in a small number of projects (e.g., lots of interest in package development, but only two package projects).
• +1 to smaller groups but I do think that the distributed organization model is also important to maintain to distinguish the session from more formal instructional sessions.

[gcperk]

Add my two cents on top!

[ateucher]

Same!

[boshek]

A suggestion for future events drawn from the evaluations: 5 minute lightning talks from participants about some real use cases of R.