epdiff: Explain difference between predictions from a model

In CollinErickson/ExplainPrediction2: Explain Difference in Predictions From a Model

Description

Explain difference between predictions from a model

Usage

epdiff(mod, x1, x2, d, predictfunc)

Arguments

mod	Model
x1	Input 1. Single row data frame or matrix.
x2	Input 2. Single row data frame or matrix.
d	Max number of input dimensions to use.
predictfunc	Func to use to predict. Defaults to 'predict'.

Value

ggplot object showing effect

Examples

# gbm
m1 <- gbm::gbm(Petal.Width ~ Sepal.Length + Sepal.Width, data=iris, distribution='gaussian')
x1 <- iris[1,]
x2 <- iris[2,]
epdiff(m1, x1, x2)
epdiff(m1, iris[12,], iris[20,])
epdiff(m1, iris[sample(1:nrow(iris), 1),], iris[sample(1:nrow(iris), 1),])

# Linear model
mod_lm <- lm(Petal.Width ~ Sepal.Length + Sepal.Width, data=iris)
epdiff(mod_lm, iris[sample(1:nrow(iris), 1),], iris[sample(1:nrow(iris), 1),])
epdiff(mod_lm, iris[sample(1:nrow(iris), 1),], iris[sample(1:nrow(iris), 1),], d=3)
mod_lm2 <- lm(Petal.Width ~ Petal.Length + Sepal.Width, data=iris)
epdiff(mod_lm2, iris[sample(1:nrow(iris), 1),], iris[sample(1:nrow(iris), 1),])

CollinErickson/ExplainPrediction2 documentation built on Dec. 17, 2021, 3:02 p.m.