Description Usage Arguments Details Value Author(s) References See Also Examples

Simulate data from a mixture of regressions model, as specified by the user or as fitted to a data set. The simulation may be done either in a parametric or “semiparametric” manner.

1 2 3 4 5 6 7 |

`x` |
For the default method, this is a numeric vector constituting
a predictor for a regression model, or a matrix whose columns
form such predictors. The number of columns of Data from a scalar mixture model may also be generated by
specifying For the |

`nobs` |
Integer scalar, specifying the number of observations to be
generated. Used only if argument |

`theta` |
Either a list or a matrix specifying the parameters of the model
from which data are to be simulated. If it is a list it should
have components |

`seed` |
A numeric scalar. If not an integer it gets rounded to the nearest
integer (so |

`xNms` |
Character vector of names for the predictors |

`yNm` |
Character scalar; a name for the response. |

`semiPar` |
Logical scalar. Should the simulation be done “semiparametrically”?
(See |

`conditional` |
Logical scalar; should the component-selection probabilities be determined conditionally upon the observations? |

`...` |
Not used. |

In this context “parametric” bootstrapping means that the
bootstrap data sets are generated by simulating from the fitted
`ncomp`

model parameters, using Gaussian errors. In contrast
semiparametric bootstrapping means that the errors are generated by
resampling from the residuals. Since at each predictor value there
are `ncomp`

residuals, one for each component of the model,
the errors are selected from these `ncomp`

possibilities.
If the argument `conditional`

is `TRUE`

then the selection
probabilities at this step are the conditional probabilities, of
the observation being generated by each component of the model,
given that observation. If `conditional`

is `FALSE`

then
these probabilities are the corresponding entries of `lambda`

(see **Value**. The residuals are sampled independently
in either case. The procedure is termed *semi*parametric
(rather than non-parametric) since the sampling probabilities depend
on the parameters of the model. Note that it makes no sense to
specify `conditional=TRUE`

if `semiPar`

is `FALSE`

.
Doing so will generate an error.

A data frame whose columns consist of the predictors and the
simulated response. For the default method the predictor are the
columns of the matrix specified by argument `x`

. They have
names given by argument `xNms`

if this was provided and by
`X1`

, `X2`

, ..., `Xn`

(where `n`

is the
number of columns of `x`

) or simply `x`

if there is
only a single predictor. For the `"mixreg"`

method the
columns are the same as those of `x$data`

, with response
column replaced by the simulated response.

Rolf Turner r.turner@auckland.ac.nz

Turner, T. R. (2000) Estimating the rate of spread of a
viral infection of potato plants via mixtures of regressions.
*Applied Statistics* **49**, Part 3 pp. 371 – 384.

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 | ```
fit <- mixreg(plntsInf ~ aphRel, ncomp=2, data=aphids)
sim1 <- rmixreg(fit)
with(sim1,plot(aphRel,plntsInf,,main="Parametric simulation"))
sim2 <- rmixreg(fit,semiPar=TRUE)
with(sim2,plot(aphRel,plntsInf,,main="Semiparametric simulation"))
x <- cbind(1:50,rnorm(50))
pmat <- matrix(c(3,5,0.01,1600,0.7,1,2,-0.01,100,0.3),nrow=2,byrow=TRUE)
sim3 <- rmixreg(x,theta=pmat,seed=42)
with(sim3,plot(X1,y,main="Using rmixreg.default; predictor 1"))
with(sim3,plot(X2,y,main="Using rmixreg.default; predictor 2"))
pmat <- matrix(c(10,2,0.7,3,1,0.3),nrow=2,byrow=TRUE)
sim4 <- rmixreg(x=rep(1,50),theta=pmat,seed=17)
sim5 <- rmixreg(x=NULL,nobs=50,theta=pmat,seed=17) # Same as sim4 but
# with no columns of 1s.
chk4 <- mixreg(y~1,data=sim4,ncomp=2,seed=116)
chk5 <- mixreg(y~1,data=sim5,ncomp=2,seed=116) # Same-same.
``` |

Embedding an R snippet on your website

Add the following code to your website.

For more information on customizing the embed code, read Embedding Snippets.