plotSPC fudge factors

[dylanbeaudette]

The extra x and y space fudge factors should be dynamically adjusted based on, number of profiles and max depth. The current approach is rather crude and based on tinkering.

# padding along x-axis, prevents crowding
  # dynamic adjustment must also taking into account figure size
  # roughly 10% of length(x)
  extra_x_space <- length(x) * 0.1

  # padding above profiles, ~ 15 is about right for n in {1,25} and max depth near 150cm
  # a sketch of shalllow profiles could benefit from ~ 5
  if(max.depth <=50)
    extra_y_space <- 5
  if(max.depth > 50 & max.depth <= 100)
    extra_y_space <- 10
  if(max.depth > 100)
    extra_y_space <- 15

This is reasonable, but we can do better with a proper scaling function.

library(aqp)
d <- lapply(1:16, random_profile, n=c(6, 7, 8), n_prop=1, method='LPP')
d <- do.call('rbind', d)
depths(d) <- id ~ top + bottom
explainPlotSPC(d)

library(aqp)
data(sp4)
depths(sp4) <- id ~ top + bottom
explainPlotSPC(d)

[dylanbeaudette]

Note that the current scaling logic doesn't work well when n < 4, and output is not acceptable when n <=2.

[dylanbeaudette]

A reasonable work-around for n = 1 and tall-narrow output.

Move sketch over 0.5 index units, make room for 2 profiles, adjust profile width accordingly.

library(soilDB)
x <- fetchOSD(c('amador', 'pentz', 'auburn', 'hornitos'))
# trick plotSPC into a reasonable figure
explainPlotSPC(x[1, ], x.idx.offset=0.5, n=2, width=0.35)

For production use you could get away with an even stranger hack:

par(mar=c(0,0,1,2))
plot(x[1, ], x.idx.offset=0.5, n=2, width=0.4, cex.names=1, y.offset=-10, scaling.factor=1.15)

I still think we need a better option for n == 1 and n<4. This is #59. Started work in R/sketch.R.