Allow users to customize `gt_plt_summary()`

[juleswg23]

• Allow for hiding of mean, median, and SD columns.
• Allow the optional addition of a Mode column
• Allow for disabling interactivity
• Allow changing colors
• Display times instead of just dates

[ScottFB101]

@juleswg23 Prior to attempting any changes, I want to confirm this is referring to gt_plt_summary(), rather than gt_summary_table(), which I'm unable to find.

[juleswg23]

I want to confirm this is referring to gt_plt_summary()

Yes this is, sorry! I'll change the title

[ScottFB101]

@juleswg23 Curious on your input here. When trying to add the optional Mode column, there's the issue of a series not containing a mode, or having multiple modes. I think displaying multiple modes could be a bad idea considering a lengthy series could contain a large number of modes, which would look weird in the table output.

The default could be to return "No Mode". Maybe even "No Singular Mode" since the formers not technically true and therefore misleading.

[juleswg23]

The default could be to return "No Mode" in that case, maybe even "No Singular Mode" since the formers not technically true and therefore misleading.

This is a good point. I think because this will be optional (and not the default) it gives us some liberties. If the user passes data where a mode is not very descriptive (continuous data, for example), and they choose to render the mode, we can't be expected to return something pretty.

For that reason, I think "No Singular Mode" works fine, or alternatively we could even check for exactly 2 modes and display both in that case in particular.

I'll leave the exact output up to you.

[ScottFB101]

Hey @juleswg23, almost ready to make a PR, but a bit confused on the addition of times to date values. Right now it doesn't seem like there's functionality to display dates.

What do you imagine the desired output of the date row to be?

[juleswg23]

The first row in the table at this link displays the current date implementation. I think it would be nice to handle times, but I didn't want to get in the weeds of datetime objects when initially writing the function.

https://posit-dev.github.io/gt-extras/examples/with-code.html#gt_plt_summary

Right now it is probably not even robust to different kinds of dates - ie numpy dates. I can check a little more carefully tomorrow.

And I will say that of the checkboxes in the initial issue, the PR doesn't have to hit all them. It is more of a roadmap of possible next steps with gt_plt_summary().

[ScottFB101]

Ah it looks like the sp500 dataset from great-tables has the dates in the string format, so _plot_categorical() ends up getting feed those values.

I'm happy to clean up what I have now and propose a PR without the date changes.