plot_effects {sgboost}

Visualizing a sparse-group boosting model

Description

Radar or scatter/lineplot visualizing the effects sizes relative to the variable importance in a sparse-group boosting model. Works also for a regular mboost model.

Usage

plot_effects(
  sgb_model,
  plot_type = "radar",
  prop = 0,
  n_predictors = 30,
  max_char_length = 5,
  base_size = 8
)

Arguments

sgb_model

Model of type mboost to be used.

plot_type

String indicating the type of visualization to use. 'radar' refers to a radar plot using polar coordinates. Here the angle is relative to the cumulative relative importance of predictors and the radius is proportional to the effect size. "clock" does the same as "radar" but uses clock coordinates instead of polar coordinates. "scatter" uses the effect size as y-coordinate and the cumulative relative importance as x-axis in a classical Scatter plot.

prop

Numeric value indicating the minimal importance a predictor/baselearner has to have to be plotted. Default value is zero, meaning all predictors are plotted. By increasing prop the number of plotted variables can be reduced. One can also use n_predictors for limiting the number of variables to be plotted directly.

n_predictors

The maximum number of predictors to be plotted. Default is 30. Alternative to prop.

max_char_length

The maximum character length of a predictor to be printed. Default is 5. For long variable names one may adjust this number.

base_size

The base_size argument to be passed to the ggplot2 theme ggplot2::theme_classic to be used to control the overall size of the figure. Default value is 8.

Value

ggplot2 object mapping the effect sizes and variable importance.

See Also

get_coef(), get_varimp() which this function uses.

Examples

library(mboost)
library(dplyr)
set.seed(1)
df <- data.frame(
  x1 = rnorm(100), x2 = rnorm(100), x3 = rnorm(100),
  x4 = rnorm(100), x5 = runif(100)
)
df <- df %>%
  mutate_all(function(x) {
    as.numeric(scale(x))
  })
df$y <- df$x1 + df$x4 + df$x5
group_df <- data.frame(
  group_name = c(1, 1, 1, 2, 2),
  var_name = c("x1", "x2", "x3", "x4", "x5")
)

sgb_formula <- as.formula(create_formula(alpha = 0.3, group_df = group_df))
sgb_model <- mboost(formula = sgb_formula, data = df)
plot_effects(sgb_model)