Upload new datasets from the app

[vrodriguezf]

trafficstars

Things to consider:

• The app should call the notebook 01_create_artifact to register the user dataset in wandb. This can be done with papermill
• Since the app is going to get bigger, having everything in just two files (server.R and ui.R) is not going to scale well. Instead, we should use shiny modules. See more here
• We can use a shiny navbar element for this module, or a shinydashboard tab if we move to shinydashboard.
• For now we can assume that the data comes as a table (rows=timesteps, columns=variables) in a csv-like file, and that timesteps are evenly spaced.
• It would be very handy to show the data with the same widget that is used in the embeddings projector, in order to select the intervals for training/validation/test
• Explore the "import dataset" Rstudio button and see if it can be called from code

[vrodriguezf]

https://projector.tensorflow.org/

[vrodriguezf]

It sould be handy to have an option here to have the validation set = train set. This can be useful in cases in which the tool wants to be used to analyse the whole dataset at hand, without predictive or future purposes.

[vrodriguezf]

Have a look to https://dreamrs.github.io/datamods

There are modules to import and validate dataset in shiny easily