fit_model

Fits generalized linear models and Cox proportional hazard models.

Multivariable fractional polynomial algorithm simultaneously selects variables and functional forms in both generalized linear models and Cox proportional hazard models. Key references for this algorithm are Royston and Altman (1994)<doi:10.2307/2986270> and Sauerbrei and Royston (2008, ISBN:978-0-470-02842-1). In addition, it can model a 'sigmoid' relationship between variable x and an outcome variable y using the approximate cumulative distribution transformation proposed by Royston (2014) <doi:10.1177/1536867X1401400206>. This feature distinguishes it from a standard fractional polynomial function, which lacks the ability to achieve such modeling.

Edwin Kipruto

mfp2

Multivariable Fractional Polynomial Models with Extensions

Michael Kammer

Patrick Royston

Willi Sauerbrei

fit_model function

<dl><dt>x</dt>
<dd>a matrix of predictors (excluding intercept) with column names.
If column names are not provided they are set according to
<code>colnames(x, do.NULL = FALSE)</code>.</dd>
<dt>y</dt>
<dd>a vector for the outcome variable for glms, and a <code>Surv</code> object
for Cox models.</dd>
<dt>family</dt>
<dd>a character strong specifying glm family to be used, or "cox"
for Cox models. The default family is set to 'Gaussian'.</dd>
<dt>method</dt>
<dd>a character string specifying the method for tie handling.
See <code><a href="/link/survival%3A%3Acoxph()?package=mfp2&version=1.0.0" data-mini-rdoc="mfp2::survival::coxph()">survival::coxph()</a></code>.</dd>
<dt>strata, control, weights, offset, rownames, nocenter</dt>
<dd>parameters for Cox
or glm. See <code><a href="/link/survival%3A%3Acoxph()?package=mfp2&version=1.0.0" data-mini-rdoc="mfp2::survival::coxph()">survival::coxph()</a></code> or <code><a href="/link/stats%3A%3Aglm()?package=mfp2&version=1.0.0" data-mini-rdoc="mfp2::stats::glm()">stats::glm()</a></code> for details.</dd>
<dt>fast</dt>
<dd>passed to <code>fit_glm()</code> and <code>fit_cox()</code>.
@return
A list with the following components:<ul>
<li><code>logl</code>: the log likelihood of the fitted model.</li>
<li><code>coefficients</code>: regression coefficients.</li>
<li><code>df</code>: number of parameters (degrees of freedom).</li>
<li><code>sse</code>: residual sum of squares.</li>
<li><code>fit</code>: the object returned by the fitting procedure.</li>
</ul></dd></dl>

Arguments

Function that fits models supported by mfp2 — fit_model

<dl>

<dt>x</dt>
<dd>a matrix of predictors (excluding intercept) with column names.
If column names are not provided they are set according to
<code>colnames(x, do.NULL = FALSE)</code>.</dd>


<dt>y</dt>
<dd>a vector for the outcome variable for glms, and a <code>Surv</code> object
for Cox models.</dd>


<dt>family</dt>
<dd>a character strong specifying glm family to be used, or "cox"
for Cox models. The default family is set to 'Gaussian'.</dd>


<dt>method</dt>
<dd>a character string specifying the method for tie handling.
See <code><a href='https://rdrr.io/pkg/survival/man/coxph.html'>survival::coxph()</a></code>.</dd>


<dt>strata, control, weights, offset, rownames, nocenter</dt>
<dd>parameters for Cox
or glm. See <code><a href='https://rdrr.io/pkg/survival/man/coxph.html'>survival::coxph()</a></code> or <code><a href='https://rdrr.io/r/stats/glm.html'>stats::glm()</a></code> for details.</dd>


<dt>fast</dt>
<dd>passed to <code>fit_glm()</code> and <code>fit_cox()</code>.
@return
A list with the following components:<ul>
<li><code>logl</code>: the log likelihood of the fitted model.</li>
<li><code>coefficients</code>: regression coefficients.</li>
<li><code>df</code>: number of parameters (degrees of freedom).</li>
<li><code>sse</code>: residual sum of squares.</li>
<li><code>fit</code>: the object returned by the fitting procedure.</li>
</ul></dd>

</dl>

fit_model: Function that fits models supported by `mfp2`

Description

Usage

Arguments

Details