interactionPlot: interactionPlot

In AlanInglis/vividOld: Variable Importance and Variable Interaction Displays

Description

Plots a selected variable against all other variables in a model

Usage

interactionPlot(
  model,
  data,
  interactionType = "ice",
  pal = rev(sequential_hcl(palette = "Blues 3", n = 11)),
  fitlims = NULL,
  gridSize = 10,
  normalize = FALSE,
  n_max = 1000,
  seed = NULL,
  sqrt = FALSE,
  type = "lollipop",
  title = NULL,
  label = FALSE
)

Arguments

model	A machine learning model created from mlr3 task and learner.
interactionType	Are measures based on Friedman's H statistic ("H") or on "ice" curves?
pal	A vector of colors to show values, for use with scale_fill_gradientn
fitlims	Specifies the fit range for the color map for interaction strength.
gridSize	The size of the grid for evaluating the predictions.
normalize	Should the variances explained be normalized? Default is FALSE.
n_max	Maximum number of data rows to consider.
seed	An integer random seed used for subsampling.
sqrt	In order to reproduce Friedman's H statistic, resulting values are root transformed. Set to FALSE if squared values should be returned.
type	The type of plot to display, either "lollipop" (default) or "barplot".
title	A title to be placed on the plot
label	Only compatible with plotType = "barplot". If TRUE then the interaction value is displayed at the end of each bar.

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 plot:
interactionPlot(aq_Task, aq_Mod)

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