Model Selection Based on Combined Penalties. This package implements a stepwise forward variable selection algorithm based on a penalized likelihood criterion that combines the L0 with L2 or L1 norms.

This package provides a graphical user interface (GUI) for fitting Bayesian regression models using the package brms which in turn relies on Stan (<https://mc-stan.org/>). The shinybrms GUI is a shiny app.

Implementations of stochastic, limited-memory quasi-Newton optimizers, similar in spirit to the LBFGS (Limited-memory Broyden-Fletcher-Goldfarb-Shanno) algorithm, for smooth stochastic optimization. Implements the following methods: oLBFGS (online LBFGS) (Schraudolph, N.N., Yu, J. and Guenter, S., 2007 <http://proceedings.mlr.press/v2/schraudolph07a.html>), SQN (stochastic quasi-Newton) (Byrd, R.H., Hansen, S.L., Nocedal, J. and Singer, Y., 2016 <arXiv:1401.7020>), adaQN (adaptive quasi-Newton) (Keskar, N.S., Berahas, A.S., 2016, <arXiv:1511.01169>). Provides functions for easily creating R objects with partial_fit/predict methods from some given objective/gradient/predict functions. Includes an example stochastic logistic regression using these optimizers. Provides header files and registered C routines for using it directly from C/C++.

Visualisation tools for SMITIDstruct package. Allow to visualize host timeline, transmission tree, index diversities and variant graph using HTMLwidgets'. It mainly using D3JS javascript framework.

This package provides efficient R and C++ routines to simulate cognitive diagnostic model data for Deterministic Input, Noisy "And" Gate ('DINA') and reduced Reparameterized Unified Model ('rRUM') from Culpepper and Hudson (2017) <doi: 10.1177/0146621617707511>, Culpepper (2015) <doi:10.3102/1076998615595403>, and de la Torre (2009) <doi:10.3102/1076998607309474>.

An efficient implementation of SCCI using Rcpp'. SCCI is short for the Stochastic Complexity-based Conditional Independence criterium (Marx and Vreeken, 2019). SCCI is an asymptotically unbiased and L2 consistent estimator of (conditional) mutual information for discrete data.

The predictive value of a statistical model can often be improved by applying shrinkage methods. This can be achieved, e.g., by regularized regression or empirical Bayes approaches. Various types of shrinkage factors can also be estimated after a maximum likelihood. While global shrinkage modifies all regression coefficients by the same factor, parameterwise shrinkage factors differ between regression coefficients. With variables which are either highly correlated or associated with regard to contents, such as several columns of a design matrix describing a nonlinear effect, parameterwise shrinkage factors are not interpretable and a compromise between global and parameterwise shrinkage, termed joint shrinkage', is a useful extension. A computational shortcut to resampling-based shrinkage factor estimation based on DFBETA residuals can be applied. Global, parameterwise and joint shrinkage for models fitted by lm(), glm(), coxph(), or mfp() is available.

Implement a promising, and yet little explored protocol for bioacoustical analysis, the eigensound method by MacLeod, Krieger and Jones (2013) <doi:10.4404/hystrix-24.1-6299>. Eigensound is a multidisciplinary method focused on the direct comparison between stereotyped sounds from different species. SoundShape', in turn, provide the tools required for anyone to go from sound waves to Principal Components Analysis, using tools extracted from traditional bioacoustics (i.e. tuneR and seewave packages), geometric morphometrics (i.e. geomorph package) and multivariate analysis (e.g. stats package). For more information, please see Rocha and Romano (2021) and check SoundShape repository on GitHub for news and updates <https://github.com/p-rocha/SoundShape>.

This package provides functions to calculate exact critical values, statistical power, expected time to signal, and required sample sizes for performing exact sequential analysis. All these calculations can be done for either Poisson or binomial data, for continuous or group sequential analyses, and for different types of rejection boundaries. In case of group sequential analyses, the group sizes do not have to be specified in advance and the alpha spending can be arbitrarily settled. For regression versions of the methods, Monte Carlo and asymptotic methods are used.

Some R functions, such as optim(), require a function its gradient passed as separate arguments. When these are expensive to calculate it may be much faster to calculate the function (fn) and gradient (gr) together since they often share many calculations (chain rule). This package allows the user to pass in a single function that returns both the function and gradient, then splits (hence splitfngr') them so the results can be accessed separately. The functions provided allow this to be done with any number of functions/values, not just for functions and gradients.

This package provides a spectral framework to map quantitative trait loci (QTLs) affecting joint differential networks of gene co-Expression. Test the equivalence among multiple biological networks via spectral statistics. See reference Hu, J., Weber, J. N., Fuess, L. E., Steinel, N. C., Bolnick, D. I., & Wang, M. (2025) <doi:10.1371/journal.pcbi.1012953>.

Efficient framework to estimate high-dimensional generalized matrix factorization models using penalized maximum likelihood under a dispersion exponential family specification. Either deterministic and stochastic methods are implemented for the numerical maximization. In particular, the package implements the stochastic gradient descent algorithm with a block-wise mini-batch strategy to speed up the computations and an efficient adaptive learning rate schedule to stabilize the convergence. All the theoretical details can be found in Castiglione et al. (2024, <doi:10.48550/arXiv.2412.20509>). Other methods considered for the optimization are the alternated iterative re-weighted least squares and the quasi-Newton method with diagonal approximation of the Fisher information matrix discussed in Kidzinski et al. (2022, <http://jmlr.org/papers/v23/20-1104.html>).

This package creates classifier for binary outcomes using Adaptive Boosting (AdaBoost) algorithm on decision stumps with a fast C++ implementation. For a description of AdaBoost, see Freund and Schapire (1997) <doi:10.1006/jcss.1997.1504>. This type of classifier is nonlinear, but easy to interpret and visualize. Feature vectors may be a combination of continuous (numeric) and categorical (string, factor) elements. Methods for classifier assessment, predictions, and cross-validation also included.

These are tools that allow users to do time series diagnostics, primarily tests of unit root, by way of simulation. While there is nothing necessarily wrong with the received wisdom of critical values generated decades ago, simulation provides its own perks. Not only is simulation broadly informative as to what these various test statistics do and what are their plausible values, simulation provides more flexibility for assessing unit root by way of different thresholds or different hypothesized distributions.

Estimates Hessian of a scalar-valued function, and returns it in a sparse Matrix format. The sparsity pattern must be known in advance. The algorithm is especially efficient for hierarchical models with a large number of heterogeneous units. See Braun, M. (2017) <doi:10.18637/jss.v082.i10>.

This package contains various functions to be used for simulation education, including simple Monte Carlo simulation functions, queueing simulation functions, variate generation functions capable of producing independent streams and antithetic variates, functions for illustrating random variate generation for various discrete and continuous distributions, and functions to compute time-persistent statistics. Also contains functions for visualizing: event-driven details of a single-server queue model; a Lehmer random number generator; variate generation via acceptance-rejection; and of generating a non-homogeneous Poisson process via thinning. Also contains two queueing data sets (one fabricated, one real-world) to facilitate input modeling. More details on the use of these functions can be found in Lawson and Leemis (2015) <doi:10.1109/WSC.2017.8248124>, in Kudlay, Lawson, and Leemis (2020) <doi:10.1109/WSC48552.2020.9384010>, and in Lawson and Leemis (2021) <doi:10.1109/WSC52266.2021.9715299>.

This package provides a novel semi-supervised machine learning algorithm to predict phenotype event times using Electronic Health Record (EHR) data.

Scrap speech text and speaker informations of speeches of House of Representatives of Brazil, and transform in a cleaned tibble.

This package implements the Sliding Window Discrete Fourier Transform (SWDFT). Also provides statistical methods based on the SWDFT, and graphical tools to display the outputs.

Package provides the possibility to sampling complete datasets from a normal distribution to simulate cluster randomized trails for different study designs.

Add significance marks to any R Boxplot, including a given significance niveau.

Insert Glide JavaScript component into Shiny applications for carousel or assistant-like user interfaces.

For Multi Parent Populations (MPP) Identity By Descend (IBD) probabilities are computed using Hidden Markov Models. These probabilities are then used in a mixed model approach for QTL Mapping as described in Li et al. (<doi:10.1007/s00122-021-03919-7>).

Provide estimation and data generation tools for the skew-unit family discussed based on Mukhopadhyay and Brani (1995) <doi:10.2307/2348710>. The family contains extensions for popular distributions such as the ArcSin discussed in Arnold and Groeneveld (1980) <doi:10.1080/01621459.1980.10477449>, triangular, U-quadratic and Johnson-SB proposed in Cortina-Borja (2006) <doi:10.1111/j.1467-985X.2006.00446_12.x> distributions, among others.