AlphaSensAnalysis: Sensitivity alpha-curve associated to MLP function

In NeuralSens: Sensitivity Analysis of Neural Networks

Sensitivity alpha-curve associated to MLP function

Description

Obtain sensitivity alpha-curves associated to MLP function obtained from the sensitivities returned by SensAnalysisMLP.

Usage

AlphaSensAnalysis(
  sens,
  tol = NULL,
  max_alpha = 15,
  interpolate_alpha = FALSE,
  curve_equal_length = FALSE,
  curve_equal_origin = FALSE,
  curve_divided_max = FALSE,
  inp_var = NULL,
  line_width = 1
)

Arguments

sens	sensitivity object returned by SensAnalysisMLP
tol	difference between M_alpha and maximum sensitivity of the sensitivity of each input variable
max_alpha	maximum alpha value to analyze
interpolate_alpha	interpolate alpha mean if difference of maximum sensitivity and last alpha evaluated is less than tol
curve_equal_length	make all the curves of the same length
curve_equal_origin	make all the curves begin at (1,0)
curve_divided_max	create second plot of curves divided by maximum alpha
inp_var	character indicating which input variable to show in density plot. Only useful when choosing plot_type='raw' to show the density plot of one input variable. If NULL, all variables are plotted in density plot. By default is NULL.
line_width	int width of the line in the plot.

Value

alpha-curves of the MLP function

Examples

mod <- RSNNS::mlp(simdata[, c("X1", "X2", "X3")], simdata[, "Y"],
                 maxit = 1000, size = 15, linOut = TRUE)

sens <- SensAnalysisMLP(mod, trData = simdata,
                        output_name = "Y", plot = FALSE)

AlphaSensAnalysis(sens)

NeuralSens documentation built on July 9, 2023, 6:18 p.m.