Skip to contents

Plot LHWR Objects

Source: R/plot.R
plot.lhwr.Rd

Plot local model outputs from a fitted LHWR model or plot a bandwidth selection curve. Model plots can show local coefficients, standard errors, local t-values, local p-values, local significance flags, fitted values, residuals, or local R-squared values. The same quantity can be drawn either on the projected (z1, z2) coordinates or on a perspective view of the Lorentz hyperboloid.

Usage

# S3 method for class 'lhwr'
plot(
  x,
  type = c("coefficient", "se", "t_value", "p_value", "p_adjusted", "significance",
    "residual", "fitted", "local_r2"),
  coefficient = NULL,
  Z = NULL,
  geometry = c("projection", "hyperboloid"),
  palette = "Viridis",
  point_size = 1.1,
  main = NULL,
  inference_df = c("edf", "residual", "normal"),
  p_adjust = c("BH", "none", "holm", "bonferroni", "BY", "fdr"),
  alpha = 0.05,
  ...
)

Arguments

x

An object returned by lhwr() or bw.lhwr().

type

Quantity to plot for model objects. "coefficient" plots a local coefficient estimate. "se" plots its local standard error. "t_value" plots its local t-value. "p_value" and "p_adjusted" plot local inference p-values. "significance" plots a 0/1 local significance flag based on adjusted p-values. "residual", "fitted", and "local_r2" plot model diagnostics stored in the fitted object.

coefficient

Coefficient name to plot for coefficient-level plot types. Defaults to the first local coefficient.

Z

Optional n x 3 Lorentz coordinate matrix. If omitted for model objects, the coordinates stored in the fitted object are used.

geometry

Plotting geometry. "projection" draws a fast two-dimensional map using (z1, z2). "hyperboloid" draws the upper Lorentz hyperboloid surface x0 = sqrt(1 + x1^2 + x2^2) and overlays the observations.

palette

Color palette passed to grDevices::hcl.colors().

point_size

Point size.

main

Optional plot title.

inference_df

Degrees-of-freedom rule used when plotting local p-values; passed to local_inference().

p_adjust

Multiplicity adjustment used when plotting adjusted p-values or significance; passed to local_inference().

alpha

Significance level used when type = "significance".

...

Additional arguments passed to base plotting functions.

Value

Invisibly returns x.

Examples

set.seed(4)
H <- simulate_H2(60, r_max = 3)
Z <- H$Z
x1 <- rnorm(60)
y <- 1 + 2 * x1 + rnorm(60, sd = 0.3)
dat <- data.frame(y = y, x1 = x1)
fit <- lhwr(y ~ x1, dat, Z = Z, bw = 30, adaptive = TRUE, diagnostics = "full")
plot(fit, coefficient = "x1")
plot(fit, type = "p_value", coefficient = "x1")
plot(fit, type = "significance", coefficient = "x1")
plot(fit, type = "residual")
plot(fit, coefficient = "x1", geometry = "hyperboloid")