vividMatrix: vividMatrix

In AlanInglis/vividOld: Variable Importance and Variable Interaction Displays

Description

Creates a matrix displaying Variable importance on the diagonal and Variable Interaction on the off-diagonal.

Usage

vividMatrix(
  task,
  model,
  filter = NULL,
  gridSize = 10,
  normalize = FALSE,
  nmax = 500,
  sqrt = TRUE,
  reorder = TRUE,
  main = NULL,
  ...
)

Arguments

task	Task created from the mlr3 package, either regression or classification.
model	A machine learning model created from mlr3 task and learner.
gridSize	The size of the grid for evaluating the predictions.
normalize	Should the variances explained be normalized? Default is FALSE.
nmax	Maximum number of data rows to consider.
sqrt	In order to reproduce Friedman's H statistic, resulting values are root transformed. Set to FALSE if squared values should be returned.
reorder	If TRUE (default) uses DendSer to reorder the matrix of interactions and variable importances.
main	Define main category for classification.

Value

A matrix of values

Examples

# Load in the data:
aq <- data.frame(airquality)
aq <- na.omit(aq)

# Run an mlr ranger model:
library(mlr3)
library(mlr3learners)
library(ranger)
aq_Task <- TaskRegr$new(id = "airQ", backend = aq, target = "Ozone")
aq_lrn <- lrn("regr.ranger", importance = "permutation")
aq_Mod <- aq_lrn$train(aq_Task)

# Create matrix
myMat <- vividMatrix(task = aq_Task, model = aq_Mod)

# Create graph:
plot(myMat, type = "heatMap")

AlanInglis/vividOld documentation built on March 4, 2021, 12:44 a.m.