Read on Twitter
A few months ago I spent some time rebuilding my training load shiny app for processing raw accelerometry data.
Here’s a thread going through some changes I’ve made. #rstats #shiny (1/9)
Here’s a thread going through some changes I’ve made. #rstats #shiny (1/9)
I was keen to redesign the UI so it didn’t look like every other shiny app using shiny dashboard. I took some initial design inspiration from @grrrck’s 2019 shiny contest winning app ( https://www.garrickadenbuie.com ).
Here’s the before and after. (2/9)
Here’s the before and after. (2/9)
The UI is built with shinydashboard, dashboardthemes, shinydashboardPlus, shinycssloaders, dygraphs, shinyFiles, and some poorly written HTML.
This thread by @jfernandez__ is a goldmine for more good UI additions: https://twitter.com/jfernandez__/status/1197909220890480640 (3/9)
This thread by @jfernandez__ is a goldmine for more good UI additions: https://twitter.com/jfernandez__/status/1197909220890480640 (3/9)
The timeseries graph is built using dygraphs, which deals with the large volume of data well. I downsampled the dataset being visualised from 100 Hz to 2 Hz to speed up the loading time. Also added loading widgets 
shinycssloaders::withSpinner (4/9)
shinycssloaders::withSpinner (4/9)
Dygraphs has great interactivity. In particular, I’ve used the dateWindow option to subset the data by selecting an area of the graph, making it really easy to clip sessions. The value boxes are linked to the subsetted data, so give some instant simple feedback on a session (5/9)
First time round I used fileInput to select the raw data, but found this to be pretty slow at uploading the file. This time I’ve used a combination of shinyFiles and data.table::fread, which is much quicker. The shinyFiles interface is also better IMO. (6/9)
Previously, for each day the app would export summary data to a csv, and then I’d copy paste into a csv database. I’ve taken @Mladen_Sormaz’s advice, and moved the database to SQL. Using DBI::dbConnect I can now add to/remove from/edit the database directly through shiny. (7/9)
At some point I’ll hopefully get round to integrating the @smartabase API, and upload data to our athlete management system directly from the shiny app. (8/9)
I took some time to clean up the script, and label/section the code so it doesn’t take forever to navigate. It’s kinda worrying how many lines I had devoted to creating a graveyard for forgotten ggplots. I’m going to assume that everyone does this 
(9/9)
(9/9)
