Generic negative log-likelihood for transformation models

Allows for the specification of deep conditional transformation
models (DCTMs) and ordinal neural network transformation models, as
described in Baumann et al (2021) <doi:10.1007/978-3-030-86523-8_1> and
Kook et al (2022) <doi:10.1016/j.patcog.2021.108263>. Extensions such as
autoregressive DCTMs (Ruegamer et al, 2023, <doi:10.1007/s11222-023-10212-8>)
and transformation ensembles (Kook et al, 2022, <doi:10.48550/arXiv.2205.12729>)
are implemented. The software package is described in Kook et al (2024,
<doi:10.18637/jss.v111.i10>).

Lucas Kook

deeptrafo

Fitting Deep Conditional Transformation Models

Philipp Baumann

David Ruegamer

nll function

<dl><dt>latent_distr</dt>
<dd>Target distribution, character or
<code>tfd_distribution</code>. If character, can be either "logistic",
"normal", "gumbel", "gompertz".</dd></dl>

Arguments

Generic negative log-likelihood for transformation models — nll

<dl>

<dt>latent_distr</dt>
<dd>Target distribution, character or
<code>tfd_distribution</code>. If character, can be either "logistic",
"normal", "gumbel", "gompertz".</dd>

</dl>

nll: Generic negative log-likelihood for transformation models

Description

Usage

Value

Arguments