lmIntXY: Confidence and prediction intervals for regression.

In tsieger/tsiMisc: Miscelaneous utility functions, mostly related to exploratory statistical analysis

Confidence and prediction intervals for regression.

Description

lmInt contructs confidence and prediction intervals for simple linear regression. For a simple univariate linear model in the form of EY = beta_0 + beta_1 X computed using 'lm(y ~ x)', 'lmIntervals' computes confidence intervals around the regression line (i.e. the point-wise confidence bands of E(Y | X = \code{x}) for each individual x), confidence intervals for the regression line (i.e. the simultaneous confidence bands of E(Y) for all x), and prediction intervals (i.e. point-wise confidence bands for new observations Y for each individual x).

Usage

lmIntXY(x, y, d = 100)

Arguments

x	independent variable, or a fitted linear model
y	dependent variable
d	a vector of values to compute the intervals for, or a number of values to be automatically generated to uniformly span the range of ‘x’

Details

The confidence intervals around the regression line and the prediction intervals are computed using the 'predict.lm' function. The confidence intervals for the regression line are computed according to eq. (4.15) in Zvara2008.

Value

An object of class lmInt - a data frame of columns x holding the values of the independent variable the intervals are computed for, fit holding the mean value fitted by the regression model, ciaLwr and ciaUpr holding confidence intervals around the regression line, cifLwr and cifUpr holding confidence intervals for the regression line, and piLwr and piUpr holding predictions intervals.

Note

deprecated, use lmInt instead

Author(s)

Tomas Sieger

References

Karel Zv\'ara: Regrese, Matfyzpress Praha 2008

See Also

predict.lm, lm

Examples

iris.setosa<-iris[iris$Species=='setosa',]
attach(iris.setosa)
lmi <- lmIntXY(Sepal.Length, Sepal.Width, 30)
plot(Sepal.Length, Sepal.Width)
plot(lmi, fit = TRUE, lty = 1, col='red')
plot(lmi, cia = TRUE, lty = 1)
plot(lmi, cif = TRUE, lty = 2)
plot(lmi, pi = TRUE, lty = 3)
legend('topright', bg='white',
  c('fitted regression line',
  'confidence interval around the regression line',
  'confidence interval for the regression line',
  'prediction int.'),
  col = c('red', 'black', 'black', 'black'), lty = c(1, 1, 2, 3))
detach(iris.setosa)

tsieger/tsiMisc documentation built on Oct. 10, 2023, 10:24 p.m.