predict.linear.bayes: Residuals or Predicted Values for linear.bayes objects

In bqtl: Bayesian QTL Mapping Toolkit

Residuals or Predicted Values for linear.bayes objects

Description

The linear.bayes object returns fitted coefficients. These are used to construct predicted values. Since the fitting process for linear.bayes objects is based on moments of centered variables, the 'intercept' is lost; see ‘Details’ below.

Usage

## S3 method for class 'linear.bayes'
residuals(object, ...)
## S3 method for class 'linear.bayes'
predict(object, newdata = lb.call$ana.obj, return.resids =
                 FALSE, ...)
## S3 method for class 'linear.bayes'
fitted(...)

Arguments

object	An object returned by linear.bayes
...	possibly the following
newdata	Optional data.frame in which to do the calculations
return.resids	Not usually set by the user.

Details

Since the linear.bayes object is based on a moment matrix, some information is lost thsat must be reconstructed or assumed. The intercept and possibly the coefficients for control variates are aong these. Also, when the call to linear.bayes supplied the moment matrix rather than formulae with which to create one, then it is unclear what variable was used as the regressand and hence which variable to use in ofrming residuals. So, in that case, residuals will report an error

Value

A vector of predicted values or residuals

Author(s)

Charles C. Berry [e-mail]

See Also

linear.bayes

bqtl documentation built on Sept. 25, 2024, 1:08 a.m.