Plot conditional density curves

This function plots the model-based conditional density curves for different effect combinations. For example, take a particular covariate that is selected in the final model. The other selected covariates are fixed at their median values by default (see covariate_fix to fix at other values) and then the plotted red and blue densities correspond to the modification of the chosen covariate as “low” (25th quantile by default) and “high” (75th quantile by default).

Usage

plot_effects(
  obj,
  what = "all",
  show_average_indiv = TRUE,
  p = c(0.25, 0.75),
  covariate_fix,
  density_range
)

Arguments

obj

An object of class “smoothic” which is the result of a call to smoothic.

what

The covariate effects to be plotted, default is what = "all". The user may supply a vector of covariate names to be plotted (only covariates selected in the final model can be plotted).

show_average_indiv

Should a “baseline” or “average” individual be shown, default is show_average_indiv = TRUE. If show_average_indiv = FALSE then this is not shown.

p

The probabilities given to the quantile function. This corresponds to the plotted red and blue density curves where the chosen covariate is modified as “low” and “high”. The default is p = c(0.25, 0.75) to show the 25th and 75th quantiles.

covariate_fix

Optional values to fix the covariates at that are chosen in the final model. When not supplied, the covariates are fixed at their median values. See the example for more detail.

density_range

Optional range for which the density curves should be plotted.

Value

A plot of the conditional density curves.

Author

Meadhbh O'Neill

Examples

# Sniffer Data --------------------
# MPR Model ----
results <- smoothic(
  formula = y ~ .,
  data = sniffer,
  family = "normal",
  model = "mpr"
)
plot_effects(results)


# Only plot gastemp and gaspres
# Do not show the average individual plot
# Plot the lower and upper density curves using 10th quantile (lower) and 90th quantile (upper)
# Fix violent to its violent to 820 and funding to 40

plot_effects(results,
             what = c("gastemp", "gaspres"),
             show_average_indiv = FALSE,
             p = c(0.1, 0.9),
             covariate_fix = c("gastemp" = 70,
                               "gaspres" = 4))


# The curves for the gastemp variable are computed by fixing gaspres = 4 (as is specified
# in the input). The remaining variables that are not specified in covariate_fix are fixed
# to their median values (i.e., tanktemp is fixed at its median). gastemp is then modified
# to be low (10th quantile) and high (90th quantile), as specified by p in the function.