blm: Bayesian Linear Regression

In manschmi/blm: Bayesian Linear Regression

Description

Computes a bayesian linear regression fit.

Usage

blm(formula, prior = NULL, beta, ...)

Arguments

formula	an object of class "formula": a symbolic description of the model to be fitted.
prior	the prior distribution, either a mvnd or a blm object.
beta	the precision of the data used for the model.
...	Additional arguments and values.

Details

Models for blm are provided as for lm. If prior distribution prior of the weights is not provided the prior means are set to 0 and the variances to 1.

Value

A object of class blm. An object of class "lm" is a list containing at least the following components:

• call: the matched call

• formula: the formula used

• df.residual: the degrees of freedom of the model

• frame: the model frame used

• matrix: the model matrix used

• beta: the precision of the data

• prior: the prior distribution used

• posterior: the posterior distribution

Examples

  w0 <- 0.3 ; w1 <- 1.1 ; b <- 1.3 
  x <- rnorm(50)
  y <- rnorm(50, w1 * x + w0, 1/b) 
  mod <- blm(y~x, data=data.frame(x=x, y=y))
  mod
  plot(mod)
  
  #use with known precision
  mod_det <- blm(y~x, beta=b, data=data.frame(x=x, y=y))
  mod_det
  
  #use of a prior, typically from an existing model 
  x2 <- rnorm(50) 
  y2 <- rnorm(50, w1 * x2 + w0, 1/b) 
  mod2 <- blm(y~x, prior=mod, data=data.frame(x=x2, y=y2)) 
  mod2
  
  #use with 2 explanatory variables 
  w2 <- 3.3 
  z <- rnorm(50) 
  y <- rnorm(50, w2 * z + w1 * x + w0, 1/b) 
  mod <- blm(y~x+z, data=data.frame(x=x, y=y, z=z)) 
  mod
 

manschmi/blm documentation built on May 21, 2019, 11:25 a.m.