ggwaterfall w/ start and end balances - expanding discrete x scales problem.

[EvaMaeRey]

What to do about discrete scales markers that that don't come from the data is still something I think about all the time because of ggwaterfall.

The 'pure' waterfall probably only represents flows:

log_of_events_and_flow_amount_df |>
 ggplot() + 
     aes(x = event, flow = flow) +
     geom_waterfall(start_balance = 0)

Then you get event labels on the x axis for free. But, what's to be done about starting and end balances?

I think it could go like this:

log_of_events_and_flow_amount_df |>
 ggplot() + 
     aes(x = event, flow = flow) +
     geom_waterfall(start_balance = 200) +                           # pure waterfall that starts at 200
     geom_waterfall_balances(start_ballance = 200)            # adds bars straddling the pure waterfall - two new x positions created

Where start balance takes x position 0, and end balance takes x position max(event) + 1, and we don't get x axis label. And I think this will error because of the new x positions that don't correspond to x inputs.

I think if the required aes for geom_waterfall_balances is 'event' the error can be circumvented.

log_of_events_and_flow_amount_df |>
 ggplot() + 
     aes(x = event, flow = flow) +
     geom_waterfall(start_balance = 200) +                            # pure waterfall that starts at 200
     aes(event = event_var) +
     geom_waterfall_balances(start_ballance = 200)             # adds bars straddling the pure waterfall - two new x positions created

Then you could wrap up waterfall and waterfall_balances into something like geom_waterfall_flanked:

log_of_events_and_flow_amount_df |>
     ggplot() + 
     aes(x = event, event = event, flow = flow) +                      # inputing event twice feels not so great, but it's relatively concise I guess 
     geom_waterfall_flanked(start_balance = 200) 

No x axis labels at x = 0 and max(event) + 1 but may be worth trying out.

A scale that looks at the existing scale and modifies based on that would be nice. Really spitballing/dreamland stuff I think here.

last_plot() + 
     scale_x_discrete_addcats(cats_before = "start balance", cats_after = "end balance", 
                                                   cat_insert = "my special event", cat_at = 3)