smooth_loess: smooth parameter grid

In floatofmath/casino: Utility functions for monte carlo simulation and optimization

Description

Fits a loess regression and adds candidates scenarios within the 95 percent confidence band around the fit.

Usage

smooth_loess(scenarios, parameter, groups, points, drop = T, width = 6, ...)

Arguments

scenarios	list of candadidate parameter settings
parameter	parameter to smooth
groups	explanatory variable
points	number of points around each prediction
drop	should columns that start with "result" or "select" or be dropped (which may have been added by simulate_batch, or select_results)
width	width of the band around the smoothed fit in multiples of the standard error
...	additional options to stats::loess

Value

matrix of parameter settings with points times nrow(scenarios) rows

Author(s)

Florian Klinglmueller

Examples

scenarios <- expand.grid(m=seq(0,1,.1),s=seq(1,2,.1),n=100)
results <- simulate_batch(scenarios,100000,simulate_poly)
candidates <- select_results(results,'s',"result")
smooth_criminals <- smooth_loess(candidates,'m','s',10)

floatofmath/casino documentation built on May 16, 2019, 1:21 p.m.