Scatterplots with variables aggregated in event windows

Source: R/event.xyplot.R

Scatterplots with variables aggregated in event windows.

event.xyplot(x, ...)

# S3 method for formula
event.xyplot(
  x,
  data = list(),
  events,
  ...,
  response = NULL,
  eFUN = sum,
  as.table = TRUE,
  layout = NULL,
  auto.key = NCOL(response) > 1,
  xlab = "event-aggregated covariate values",
  ylab = "event-aggregated response"
)

Arguments

x

an object for the generic method; in the formula method this is a formula as described in the Details section. All terms in the formula should be aggregated in the given events; for this purpose a special function e() is defined as a shorthand for eventapply, with arguments e(X, FUN = sum, ...). The formula may specify a response (the left hand side of formula), which is treated as a single term, although it can be a matrix with multiple columns. Alternatively, the response matrix can be given in response. The covariate terms are also allowed to be matrix-like with the same number of columns as the response, in which case each column is plotted against the corresponsing column of the response.

...

further arguments passed to xyplot (actually xyplot.list).

data

data source giving variables used in the formula. Typically a data.frame. Passed to model.frame.

events

event sequence produced by eventseq, or a vector defining continguous groups on the specified variables.

response

a response vector or matrix corresponding to the variables in the formula. If this is given, any response in the formula is ignored.

eFUN

a function to aggregate response, if it is given, in the given events. If a reponse appears in the formula it should be in a e() term, and eFUN is not used.

as.table, layout, auto.key, xlab, ylab

passed to xyplot.

Value

this function returns a trellis object which can be plotted.

Author

Felix Andrews felix@nfrac.org

Examples


data(Canning)
ev <- eventseq(Canning$P,
  thresh = 20, inthresh = 1, indur = 3,
  continue = FALSE
)

event.xyplot(e(Q, sum) / e(P, sum) ~ e(P, sum) + e(P, max) + e(lag(Q, -1), first),
  data = Canning, events = ev,
  scales = list(y = list(log = TRUE)),
  yscale.components = yscale.components.fractions,
  ylab = "event runoff ratio (Q/P)", layout = c(3, 1),
  xlab = c(
    "total rain (mm)", "max rain (mm/day)",
    "antecedent flow (mm/day)"
  )
)


## multiple response variables
event.xyplot(cbind(e(Q, quantile, 0.9), e(Q, quantile, 0.99)) ~
e(P, max) + e(P, mean) + e(P, sum),
data = Canning, events = ev,
scales = list(y = list(log = TRUE)),
yscale.components = yscale.components.log10.3
)