README.md

In schalkdaniel/liftplot: Create a liftplot

liftplot

This package/function gives just one function called 'lift'. With this function one can descriptive validate the goodnes of the prediction of a model.

Therefore some standard classes like 'lm', 'glm', 'randomForest' or 'nls' are implemented. If a specific class isn't supported one have the opportunity to give manually a prediction vector and a vector with the real values.

Example

At first simulate some data:

set.seed( pi )
x <- sort( runif(100, 0, 2) )
y <- 3 * exp(2 * x) + rchisq(100, 10)

DF <- data.frame( x = x, y = y )

Now we can compute some models and apply lift on every one:

## For class lm:
mod1 <- lm( formula = y ~ x,
            data    = DF )

## For class nls:
mod2 <- nls( formula = y ~ a * exp(b * x),
             data    = DF,
             start   = list(a = 1, b = 1) )

## For class randomForest:
require( randomForest )
mod3 <- randomForest( formula = x ~ y,
                      data    = DF )


par( mfrow = c(1,3) )
lift(mod1, 
     col1    = c(red = 113, blue = 198, green = 113),
     col2    = c(red = 125, blue = 158, green = 192),
     bullets = "boxplot")
lift(mod2, col1 = c(red = 113, blue = 198, green = 113))
lift(mod3, col2 = c(red = 125, blue = 158, green = 192))
par( mfrow = c(1,1) )

The plot looks like the following:

schalkdaniel/liftplot documentation built on May 29, 2019, 3:26 p.m.