get_discrepancy

This function computes various discrepancy measures between observed and estimated values.
It supports different methods including logarithmic error, square error, classification error, and logistic deviance.

To improve estimation accuracy and stability in statistical modeling, catalytic prior distributions are employed, integrating observed data with synthetic data generated from a simpler model's predictive distribution. This approach enhances model robustness, stability, and flexibility in complex data scenarios. The catalytic prior distributions are introduced by 'Huang et al.' (2020, <doi:10.1073/pnas.1920913117>), Li and Huang (2023, <doi:10.48550/arXiv.2312.01411>).

Dongming Huang

catalytic

Tools for Applying Catalytic Priors in Statistical Modeling

Yitong Wu

Weihao Li

Ministry of Education, Singapore 

get_discrepancy function

<dl><dt>discrepancy_method</dt>
<dd>A character string specifying the discrepancy method to use. Options are:<dl>
<dt>"logarithmic_error"</dt>
<dd>Logarithmic error, suitable for probabilities.</dd></dl></dd><dt>"mean_square_error"</dt>
<dd>Mean squared error.</dd><dt>"mean_classification_error"</dt>
<dd>Mean of classification error, suitable for binary outcomes.</dd><dt>"logistic_deviance"</dt>
<dd>Logistic deviance, computed using a GLM model.</dd>
</dl>
<dt>family_string</dt>
<dd>A GLM family in string (e.g., "binomial") used to compute logistic deviance.</dd>
<dt>X</dt>
<dd>A matrix of predictor variables.</dd>
<dt>Y</dt>
<dd>A vector or data frame of observed values.</dd>
<dt>coefs</dt>
<dd>A vector of coefficients for the GLM model.</dd>
<dt>est_Y</dt>
<dd>A vector of estimated values. If not provided, it will be computed using <code>get_glm_mean</code> with the specified <code>family</code>.</dd>

Arguments

Compute Discrepancy Measures — get_discrepancy

<dl>

<dt>discrepancy_method</dt>
<dd>A character string specifying the discrepancy method to use. Options are:<dl>
<dt>"logarithmic_error"</dt>
<dd>Logarithmic error, suitable for probabilities.</dd>

<dt>"mean_square_error"</dt>
<dd>Mean squared error.</dd>

<dt>"mean_classification_error"</dt>
<dd>Mean of classification error, suitable for binary outcomes.</dd>

<dt>"logistic_deviance"</dt>
<dd>Logistic deviance, computed using a GLM model.</dd>


</dl></dd>


<dt>family_string</dt>
<dd>A GLM family in string (e.g., "binomial") used to compute logistic deviance.</dd>


<dt>X</dt>
<dd>A matrix of predictor variables.</dd>


<dt>Y</dt>
<dd>A vector or data frame of observed values.</dd>


<dt>coefs</dt>
<dd>A vector of coefficients for the GLM model.</dd>


<dt>est_Y</dt>
<dd>A vector of estimated values. If not provided, it will be computed using <code>get_glm_mean</code> with the specified <code>family</code>.</dd>

</dl>

get_discrepancy: Compute Discrepancy Measures

Description

Usage

Value

Arguments