inv.tran.plot: Choose a predictor transformation visually or numerically

Description

inv.tran.plot draws a two-dimensional scatterplot of $Y$ versus $X$, along with the OLS fit from the regression of $Y$ on $(X^{\lambda}-1)/\lambda$. inv.tran.estimate find the nonlinear least squares estimate of $\lambda$ and its standard error.

Usage

inv.tran.plot(x,y,lambda=c(-1,0,1),lty=1:(1+length(lambda)),
        col=rainbow(length(lambda)+1),xlab=deparse(substitute(x)),
        ylab=deparse(substitute(y)),family="box.cox",optimal=TRUE,
        key=FALSE,...)

inv.tran.estimate(x,y,family="box.cox",...)

Arguments

The predictor variable

The response variable

lambda

The powers used in the plot. The optimal power than minimizes the residual sum of squares is always added unless optimal is FALSE.

family

The transformation family to use, "box.cox", "yeo.johnson", or a user-defined family.

optimal

Include the optimal value of lambda?

lty

line types corresponding to the powers

col

color corresponding to the powers

key

If TRUE, draw a key and use locator to find it; if FALSE do not draw a key. If a vector of length 2, draw the upper left corner of the key at that point.

xlab

Label for the horizontal axis.

ylab

Label for the vertical axis.

...

additional arguments passed to other methods.

Value

inv.tran.plot returns a graph and a data.frame with $\lambda$ in the first column, and the residual sum of squares from the regression for that $\lambda$ in the second column. inv.tran.estimate returns a list with elements lambda for the estimate, se for its standard error, and RSS, the minimum value of the residual sum of squares.

References

Weisberg, S. (2005). Applied Linear Regression, third edition. New York: Wiley.

Examples

Run this code

data(baeskel)
attach(baeskel)
inv.tran.plot(Sulfur,Tension,key=c(.6,450))
ans <-inv.tran.estimate(Sulfur,Tension)
# redraw the plot, including the nls estimate
inv.tran.plot(Sulfur,Tension,lambda=c(ans$lambda,-1,0,1),key=c(.6,450))

Run the code above in your browser using DataLab

Data engineering and BI courses are free!