Profiling a Log-Likelihood to Calculate Confidence Intervals

Provides tools for profiling a user-supplied log-likelihood function to calculate confidence intervals for model parameters. Speed of computation can be improved by adjusting the step sizes in the profiling and/or starting the profiling from limits based on the approximate large sample normal distribution for the maximum likelihood estimator of a parameter. The accuracy of the limits can be set by the user. A plot method visualises the log-likelihood and confidence interval. Cases where the profile log-likelihood flattens above the value at which a confidence limit is defined can be handled, leading to a limit at plus or minus infinity. Disjoint confidence intervals will not be found.

Log-Concave Density Estimation in Arbitrary Dimensions

Software for computing a log-concave (maximum likelihood) estimator for independent and identically distributed data in any number of dimensions. For a detailed description of the method see Cule, Samworth and Stewart (2010, Journal of Royal Statistical Society Series B, ).

Bayesian Mixture Log-Normal Survival Model

Bayesian Survival models via the mixture of Log-Normal distribution extends the well-known survival models and accommodates different behaviour over time and considers higher censored survival times. The proposal combines mixture distributions Fruhwirth-Schnatter(2006) , and data augmentation techniques Tanner and Wong (1987) .

Log Binomial Regression Model in Exact Method

Fit the log binomial regression model (LBM) by Exact method. Limited parameter space of LBM causes trouble to find admissible estimates and fail to converge when MLE is close to or on the boundary of space. Exact method utilizes the property of boundary vectors to re-parametrize the model without losing any information, and fits the model on the standard fitting algorithm with no convergence issues.

Learning Sparse Log-Ratios for Compositional Data

In the context of high-throughput genetic data, CoDaCoRe identifies a set of sparse biomarkers that are predictive of a response variable of interest (Gordon-Rodriguez et al., 2021) . More generally, CoDaCoRe can be applied to any regression problem where the independent variable is Compositional (CoDa), to derive a set of scale-invariant log-ratios (ILR or SLR) that are maximally associated to a dependent variable.

Log-Multiplicative Models, Including Association Models

Functions to fit log-multiplicative models using 'gnm', with support for convenient printing, plots, and jackknife/bootstrap standard errors. For complex survey data, models can be fitted from design objects from the 'survey' package. Currently supported models include UNIDIFF (Erikson & Goldthorpe, 1992), a.k.a. log-multiplicative layer effect model (Xie, 1992) , and several association models: Goodman (1979) row-column association models of the RC(M) and RC(M)-L families with one or several dimensions; two skew-symmetric association models proposed by Yamaguchi (1990) and by van der Heijden & Mooijaart (1995) Functions allow computing the intrinsic association coefficient (see Bouchet-Valat (2022) ) and the Altham (1970) index , including via the Bayes shrinkage estimator proposed by Zhou (2015) ; and the RAS/IPF/Deming-Stephan algorithm.

Saturated Model Log-Likelihood for Multinomial Outcomes

When the response variable Y takes one of R > 1 values, the function 'glsm()' computes the maximum likelihood estimates (MLEs) of the parameters under four models: null, complete, saturated, and logistic. It also calculates the log-likelihood values for each model. This method assumes independent, non-identically distributed variables. For grouped data with a multinomial outcome, where observations are divided into J populations, the function 'glsm()' provides estimation for any number K of explanatory variables.

Compare Models with Cross-Validated Log-Likelihood

An implementation of the cross-validated difference in means (CVDM) test by Desmarais and Harden (2014) (see also Harden and Desmarais, 2011 ) and the cross-validated median fit (CVMF) test by Desmarais and Harden (2012) . These tests use leave-one-out cross-validated log-likelihoods to assist in selecting among model estimations. You can also utilize data from Golder (2010) and Joshi & Mason (2008) that are included to facilitate examples from real-world analysis.

Robust, Logged and Reproducible Iteration at Organizational Scale

Turns one-off iterative R procedures (such as for loops, lapply() or pmap() from 'purrr') into production-grade workflows by wrapping them with orthogonal, composable execution layers. Two layers are always active: structured logging with real traceback and per-case timing; and reproducibility capture, which records the R version, loaded package versions, execution environment, the exact iteration mask, and a stat-based fingerprint of every input file referenced in the mask (with a diff_inputs() helper to detect silent drift between runs). Parallel execution (built on the 'future' framework, Bengtsson (2021) ), non-blocking background jobs, and opt-in progress reporting (via 'progressr') are implemented as optional, composable layers. Further layers (error replay, content-hash input fingerprinting, content-based case identifiers) are planned and will remain composable with the default layers.

Send Log Messages to Remote 'syslog' Server

Send 'syslog' protocol messages to a remote 'syslog' server specified by host name and TCP network port.