nycflights13 | R data package containing all out-bound flights | Map library

Here's an approach with purrr::map. First, split the name column by space. Then apply a custom function to the list that is created. We can use [ to subset the vector in each list element to be the longest word. We can determine the longest word by applying nchar to each element. which.max can tell us which one is longest.

The _char version of map will return a character vector.

split every word into characters, count them using table, sort and select the 1st value.

you are getting the error as sheet and excel are not arguments in write_xlsx. To write multiple sheets to a single xlsx you want to first put them all in named list and then supply that list to write_xlsx.

a. length() returns the number of elements in an object. When the object is a data.frame (or tibble) it will return the number of columns. The L after the number means it is an integer.

b. mean() requieres a numeric vector to work. In b you are passing a tibble to the function.

c. dplyr functions are meant to receive tibbles for input and produce tibbles for output. You can pull() a column of your tibble so it becomes a vector.

Do you want this?

You can use scale_color_hue to change the labels.

Try :

Short answer, drop the runApp() at the bottom of the code.

When you click Run App, it sources the file, which in turn executes runApp(). This in turn sources the file which in turn executes runApp(). So on and so on.

When you run interactively with + , the app actually launches when it gets to the shinyApp(ui = ui, server = server) line. This function returns as Shiny app object. When running interactively, it prints and it's print method actually launches the app, pausing the code execution. When you close the app, the rest of the script continues but you've already closed the view so you don't see it. When clicking "Run App", the file is sourced without echo and therefore the shinyApp object isn't printed and only the runApp loop starts because it's looking for your file to produce the shinyApp object, not a call to run the app.

When running interactively, you actually are stopping the R execution before it gets in the loop. Without the pause from the print of the app object when using the "Run App" button, you immediately drop into the loop.

Change the end of your code to add a message and you can see:

So on this the first thing I'd like to note is that collapse is not and probably never will be a full-blown split-apply combine computing tool like dplyr or data.table. It's focus is not on optimally executing arbitrary code expressions by groups, but on providing advanced and highly efficient grouped, weighted, time-series and panel data computations through the broad range of C++ based statistical and data transformation functions it provides. I refer to the vignette on collapse and data.table for further clarity on these points as well as integration examples.

Accordingly, I think it only makes sense to translate data.table code to collapse if (1) you've come up with an arcane expression in data.table to do something complex statistical it is is not good at (such as weighted aggregation, computing quantiles or the mode by groups, lagging / differencing an irregular panel, grouped centering or linear / polynomial fitting) (2) you actually don't need the data.table object but would much rather work with vectors / matrices / data.frame's / tibbles (3) you want to write a statistical program and would much prefer standard evaluation programming over NS eval and data.table syntax or (4) collapse is indeed substantially faster for your specific application.

Now to the specific code you have provided. It mixes standard and non-standard evaluation (e.g. through the use of get()), which is something collapse is not very good at. I'll give you 3 solutions ranging from full NS eval to full standard eval base R style programming.