Plot the cumulative lift curve of a sort-order.

LiftCurvePlot(
  frame,
  xvar,
  truthVar,
  title,
  ...,
  large_count = 1000,
  include_wizard = TRUE,
  truth_target = NULL,
  model_color = "darkblue",
  wizard_color = "darkgreen"
)

Arguments

frame

data frame to get values from

xvar

name of the independent (input or model score) column in frame

truthVar

name of the dependent (output or result to be modeled) column in frame

title

title to place on plot

...

no unnamed argument, added to force named binding of later arguments.

large_count

numeric, upper bound target for number of plotting points

include_wizard

logical, if TRUE plot the ideal or wizard plot.

truth_target

if not NULL compare to this scalar value.

model_color

color for the model curve

wizard_color

color for the "wizard" (best possible) curve

Details

The use case for this visualization is to compare a predictive model score to an actual outcome (either binary (0/1) or continuous). In this case the lift curve plot measures how well the model score sorts the data compared to the true outcome value.

The x-axis represents the fraction of items seen when sorted by score, and the y-axis represents the lift seen so far (cumulative value of model over cummulative value of random selection)..

For comparison, LiftCurvePlot also plots the "wizard curve": the lift curve when the data is sorted according to its true outcome.

To improve presentation quality, the plot is limited to approximately large_count points (default: 1000). For larger data sets, the data is appropriately randomly sampled down before plotting.

Examples

set.seed(34903490) y = abs(rnorm(20)) + 0.1 x = abs(y + 0.5*rnorm(20)) frm = data.frame(model=x, value=y) WVPlots::LiftCurvePlot(frm, "model", "value", title="Example Continuous Lift Curve")