When calling easy_random_forest, variable importances scores from the
generate_variable_importances output are processed by the
process_variable_importances function and generated into
a plot. Importance scores for each predictor were estimated using the increase in
node impurity. Node impurity measures the change in residual squared error
that is attributable to the predictor across all trees. Unlike the
easy_glmnet coefficients, random forest importance scores
do not indicate directional effects, but instead represent the magnitude
of the effect that the predictor has
on overall prediction accuracy. Be careful using this plotting
method with datasets containing more than 20 variables as the plot
may not render as nicely.
plot_variable_importances_processed(variable_importances_processed)
| variable_importances_processed | A data.frame, the output of the function |
|---|
A ggplot object. This plot may be rendered by outputting it to the command line or modified using ggplot semantics.
Other plot: plot_coefficients_processed,
plot_model_performance_binomial_auc_score,
plot_model_performance_gaussian_correlation_score,
plot_model_performance_gaussian_mse_score,
plot_model_performance_gaussian_r2_score,
plot_model_performance_histogram,
plot_predictions_binomial,
plot_predictions_gaussian,
plot_roc_curve