There are probably thousands of things to do in the middle of May at the Baltic seaside, but in 2019 the place to be was Gdansk. Despite the dreadful weather, Gdańsk offered the hottest R event in Europe at the time: the very first edition of the SatRday Gdańsk.
Appsilon had the pleasure to not only be a sponsor of the whole event, but we were also invited to share our knowledge about creating Shiny Dashboards. The workshop session was held on Friday, before the main stage was opened. In parallel to our “Shiny Dashboard” course, there were two other workshops on keras and tidyverse. Even though we were competing against panels with hot topics, we managed to attract more than 20 Shiny enthusiasts. We started from scratch, by outlining the main features of Shiny like neat UI, reactivity, integration with other interactive packages like leaflet, plotly, etc. Based on that, we created a couple of Shiny Apps and a Shiny Dashboard. We also discussed the pros and cons of a standard bootstrap engine and switched to an alternative: Semantic UI. Finally, the participants of the workshop created the final project displaying in an interactive way a correlation between IMDB and Rotten Tomatoes movies’ scores with our semantic.dahboard package. All the materials from our workshop you can find at our GitHub repository here.
Naturally, the main event kicked off early Saturday morning with an introduction from the organisers. The event was held at the main building of the Gdansk University of Technology. In the morning session, Dr Agnieszka Król introduced us to the frailtypack and discussed its applications to her cancer research. The talk was full of equations and scientific content, so it was a nice warm-up for the rest of the very intense day. In general, the morning session was focused on R application to bioinformatics with talks discussing packages like predPCR or AmyloGram.
The next couple of talks were dedicated mostly to machine learning. The sessions were running in parallel in two rooms, so I couldn’t participate in all. And I regret that wholeheartedly, as I heard that both were similarly entertaining. Marcin Kosiński from Gradient, despite his initial suggestions that the other talk would be more interesting, proved that non-negative matrix factorization is not only an exciting area of statistical research, but also that it can be useful for segmentation of respondents in high dimensional surveys. I really enjoyed his theory-driven approach for building statistical models.
The talk from which I probably got the most out of was given by Dr Przemysław Biecek and Hanna Piotrowska. They convinced us of the importance of understanding Machine Learning models (for example, using package DALEX) along with how crucial is their graphical representation for conveying insights at a glance, especially for non-experts. I could elaborate about this talk even more, but instead you can check their slides out here.
Not less interesting was the talk by Dr Błażej Kochański, who described his research about modelling ROC curves. Whoever has worked with tuning ML models probably knows about the importance of this concept. The last “regular talk” of the day was given by Lubomír Štěpánek from Czech Technical University in Prague, who presented an impressive piece of work about bridging Shiny with a C++ image processing library for facial attractiveness evaluation for purposes of plastic surgery. That was followed by a quick session of three lightning talks during which I learned for example about the drake package for reproducible workflow management in R.
The conference concluded with a tour around Gdansk’s old town and a little evening gathering at the Cathead pub where we were greeted with selection of delicious dishes from traditional Polish cuisine and a couple of pints 😉 All in all, I enjoyed the SatRday in Gdansk quite a bit. A big shout out to the organizers! I can’t wait for next year’s edition!