R Shiny application for finding out about your girlfriend's engagement ring preferences.
At each round two random images are shown. The user can click on one according to his preference. With the superlike button, one can express special liking. If out of the two images both are not liked, one can select none of those.
Initially, a data frame with the image paths is created and saved in feather format. It is alimented by the contents in the rings folder. If it already exists, then it is simply loaded. Feather is a fast storage format for R (and support in Python). It is particularely useful in web applications.
There is the option to build in easter egg images in a specials folder. The probability of showing a special image set is defined as easteregg_prob in randomid().
The user inputs are written to a timestamped csv, see sessionfile.
Shiny allows to create a clickable plot or image with renderImage() and imageOutput(). User clicking on an image or plot can be used as an input. For example:
# ui
imageOutput("image1", click = "first")
# server
observeEvent(input$first, { ... })
To append to csv we use write.table().
write.table(tblheader, file = sessionfile, sep = ",", col.names = F)
with tblheader being a one-row matrix to append, and sessionfile being the timestamped csv.
The csv file can later be used for analysis of the preference.
- The images were downloaded from a jewelry seller's website using Simple mass downloader Chrome extension.
- It supports several 100s of images.
- App displays well on a Smartphone also.
- Why would one use R Shiny for this? Yes, there are a lot of frameworks that would do the job, but why not Shiny? As I would also do the user's data analysis in R later, it's a tool that is easily available and fast for an app one needs nothing but a one-use prototype.