| Line 12... |
Line 12... |
| 12 |
loglin(table, margin, start = rep(1, length(table)), fit = FALSE,
|
12 |
loglin(table, margin, start = rep(1, length(table)), fit = FALSE,
|
| 13 |
eps = 0.1, iter = 20, param = FALSE, print = TRUE)
|
13 |
eps = 0.1, iter = 20, param = FALSE, print = TRUE)
|
| 14 |
}
|
14 |
}
|
| 15 |
\description{
|
15 |
\description{
|
| 16 |
\code{loglin} is used to fit log-linear models to multidimensional
|
16 |
\code{loglin} is used to fit log-linear models to multidimensional
|
| 17 |
contingency tables by Iterative Proportional Fitting.
|
17 |
contingency tables by Iterative Proportional Fitting.
|
| 18 |
}
|
18 |
}
|
| 19 |
\arguments{
|
19 |
\arguments{
|
| 20 |
\item{table}{a contingency table to be fit, typically the output from
|
20 |
\item{table}{a contingency table to be fit, typically the output from
|
| 21 |
\code{table}.}
|
21 |
\code{table}.}
|
| 22 |
\item{margin}{a list of vectors with the marginal totals to be fit.
|
22 |
\item{margin}{a list of vectors with the marginal totals to be fit.
|
| Line 27... |
Line 27... |
| 27 |
\code{list(c(1, 2), c(1, 3))} specifies a model which contains
|
27 |
\code{list(c(1, 2), c(1, 3))} specifies a model which contains
|
| 28 |
parameters for the grand mean, each factor, and the 1-2 and 1-3
|
28 |
parameters for the grand mean, each factor, and the 1-2 and 1-3
|
| 29 |
interactions, respectively (but no 2-3 or 1-2-3 interaction), i.e.,
|
29 |
interactions, respectively (but no 2-3 or 1-2-3 interaction), i.e.,
|
| 30 |
a model where factors 2 and 3 are independent conditional on factor
|
30 |
a model where factors 2 and 3 are independent conditional on factor
|
| 31 |
1 (sometimes represented as \sQuote{[12][13]}).
|
31 |
1 (sometimes represented as \sQuote{[12][13]}).
|
| 32 |
|
32 |
|
| 33 |
The names of factors (i.e., \code{names(dimnames(table))}) may be
|
33 |
The names of factors (i.e., \code{names(dimnames(table))}) may be
|
| 34 |
used rather than numeric indices.
|
34 |
used rather than numeric indices.
|
| 35 |
}
|
35 |
}
|
| 36 |
\item{start}{a starting estimate for the fitted table. This optional
|
36 |
\item{start}{a starting estimate for the fitted table. This optional
|
| 37 |
argument is important for incomplete tables with structural zeros
|
37 |
argument is important for incomplete tables with structural zeros
|
| Line 54... |
Line 54... |
| 54 |
\item{pearson}{the Pearson test statistic (X-squared).}
|
54 |
\item{pearson}{the Pearson test statistic (X-squared).}
|
| 55 |
\item{df}{the degrees of freedom for the fitted model. There is no
|
55 |
\item{df}{the degrees of freedom for the fitted model. There is no
|
| 56 |
adjustment for structural zeros.}
|
56 |
adjustment for structural zeros.}
|
| 57 |
\item{margin}{list of the margins that were fit. Basically the same
|
57 |
\item{margin}{list of the margins that were fit. Basically the same
|
| 58 |
as the input \code{margin}, but with numbers replaced by names
|
58 |
as the input \code{margin}, but with numbers replaced by names
|
| 59 |
where possible.}
|
59 |
where possible.}
|
| 60 |
\item{fit}{An array like \code{table} containing the fitted values.
|
60 |
\item{fit}{An array like \code{table} containing the fitted values.
|
| 61 |
Only returned if \code{fit} is \code{TRUE}.}
|
61 |
Only returned if \code{fit} is \code{TRUE}.}
|
| 62 |
\item{param}{A list containing the estimated parameters of the
|
62 |
\item{param}{A list containing the estimated parameters of the
|
| 63 |
model. The \sQuote{standard} constraints of zero marginal sums
|
63 |
model. The \sQuote{standard} constraints of zero marginal sums
|
| 64 |
(e.g., zero row and column sums for a two factor parameter) are
|
64 |
(e.g., zero row and column sums for a two factor parameter) are
|
| Line 79... |
Line 79... |
| 79 |
|
79 |
|
| 80 |
Note that the IPF steps are applied to the factors in the order given
|
80 |
Note that the IPF steps are applied to the factors in the order given
|
| 81 |
in \code{margin}. Hence if the model is decomposable and the order
|
81 |
in \code{margin}. Hence if the model is decomposable and the order
|
| 82 |
given in \code{margin} is a running intersection property ordering
|
82 |
given in \code{margin} is a running intersection property ordering
|
| 83 |
then IPF will converge in one iteration.
|
83 |
then IPF will converge in one iteration.
|
| 84 |
|
84 |
|
| 85 |
Package \CRANpkg{MASS} contains \code{loglm}, a front-end to
|
85 |
Package \CRANpkg{MASS} contains \code{loglm}, a front-end to
|
| 86 |
\code{loglin} which allows the log-linear model to be specified and
|
86 |
\code{loglin} which allows the log-linear model to be specified and
|
| 87 |
fitted in a formula-based manner similar to that of other fitting
|
87 |
fitted in a formula-based manner similar to that of other fitting
|
| 88 |
functions such as \code{lm} or \code{glm}.
|
88 |
functions such as \code{lm} or \code{glm}.
|
| 89 |
}
|
89 |
}
|
The R Project SVN
R