This package provides tools for the analysis of interval-valued data, including construction, visualization, and statistical modeling. The package provides the intData class for representing interval-valued data, along with functions to aggregate microdata and to estimate parameters of latent distributions. Barycenter and covariance matrix estimation is implemented based on the Mallows distance (Oliveira et al. (2025) <doi:10.48550/arXiv.2407.05105>). Robust estimation of the symbolic covariance matrix is implemented via the Interval Minimum Covariance Determinant (IMCD) estimator, enabling outlier detection based on the robust squared Interval-Mahalanobis distance, as proposed by Loureiro et al. (2026b) <doi:10.48550/arXiv.2604.26769>. Explainable outlier detection is supported through Shapley value based decomposition of the squared robust Interval-Mahalanobis distance, allowing assessment of variable contributions to outlyingness (Loureiro et al. (2026a) <doi:10.48550/arXiv.2606.26307>). Shapley interaction indices are also implemented, along with visualization tools to support interpretation of the results.

Runs language model agents in an existing R session, where generated code can inspect and modify objects already in memory. The peter() function supports interactive chat and programmatic calls. Conversational models use tools and return session objects; typed decision models return logical, character or numeric vectors with associated model probabilities. Supports the Anthropic <https://platform.claude.com/docs/en/api/overview>, OpenAI <https://developers.openai.com/api/reference/overview> and Google Gemini <https://ai.google.dev/gemini-api/docs> APIs, OpenAI'-compatible endpoints, local Ollama <https://ollama.com> models, and the Claude Code and Codex command line tools. Decision models include Jev', hosted by TypeSafe AI <https://docs.typesafe.ai/introduction>, and the local Clef'. Top-level agent calls and their executed code can be recorded in R scripts, R Markdown and Quarto documents, or Jupyter notebooks. Recorded R scripts can be replayed without model calls when the required inputs and packages are available.

Kevin Dowd's book Measuring Market Risk is a widely read book in the area of risk measurement by students and practitioners alike. As he claims, MATLAB indeed might have been the most suitable language when he originally wrote the functions, but, with growing popularity of R it is not entirely valid. As Dowd's code was not intended to be error free and were mainly for reference, some functions in this package have inherited those errors. An attempt will be made in future releases to identify and correct them. Dowd's original code can be downloaded from www.kevindowd.org/measuring-market-risk/. It should be noted that Dowd offers both MMR2 and MMR1 toolboxes. Only MMR2 was ported to R. MMR2 is more recent version of MMR1 toolbox and they both have mostly similar function. The toolbox mainly contains different parametric and non parametric methods for measurement of market risk as well as backtesting risk measurement methods.

This package provides a comprehensive suite of genome-wide association study (GWAS) methods specifically designed for biobank-scale data, including but not limited to, robust approaches for time-to-event traits (Li et al., 2025 <doi:10.1038/s43588-025-00864-z>) and ordinal categorical traits (Bi et al., 2021 <doi:10.1016/j.ajhg.2021.03.019>). The package also offers general frameworks for GWAS of any trait type (Bi et al., 2020 <doi:10.1016/j.ajhg.2020.06.003>), while accounting for sample relatedness (Xu et al., 2025 <doi:10.1038/s41467-025-56669-1>) or population structure (Ma et al., 2025 <doi:10.1186/s13059-025-03827-9>). By accurately approximating score statistic distributions using saddlepoint approximation (SPA), these methods can effectively control type I error rates for rare variants and in the presence of unbalanced phenotype distributions. Additionally, the package includes functions for simulating genotype and phenotype data to support research and method development.

This package provides a decision support tool for prioritizing conservation projects. Prioritizations can be developed by maximizing expected feature richness, expected phylogenetic diversity, the number of features that meet persistence targets, or identifying a set of projects that meet persistence targets for minimal cost. Constraints (e.g. lock in specific actions) and feature weights can also be specified to further customize prioritizations. After defining a project prioritization problem, solutions can be obtained using exact algorithms, heuristic algorithms, or random processes. In particular, it is recommended to install the Gurobi optimizer (available from <https://www.gurobi.com>) because it can identify optimal solutions very quickly. The rcbc R package (available at <https://github.com/dirkschumacher/rcbc>) can also be used to generate solutions using the CBC optimization software (<https://github.com/coin-or/Cbc>). Finally, methods are provided for comparing different prioritizations and evaluating their benefits. For more information, see Hanson et al. (2019) <doi:10.1111/2041-210X.13264>.

Interactive spatial visualisations are a cornerstone for exploring and communicating complexity, and are commonly embedded into reports or interactive dashboards. However, as the amount of data grows, so do the demands on functionality, especially for technical and scientific data. To bridge this gap and create a mapping package that is high performing, a modern approach is needed that draws from best software engineering practices. Toro provides bindings to MapLibre GL JS', an open-source JavaScript'/'TypeScript library for rendering interactive maps in the browser, built from the ground up for responsiveness and scale, by the MapLibre Organization (2020) <https://github.com/MapLibre>. This connection allows users to create interactive maps that can easily be integrated into both Quarto and the R Shiny dashboard framework. Toro thereby enables spatial visualisation and exploration of data that might otherwise be too limited, too slow, or too hard to scale using more traditional interactive mapping tools such as leaflet'.

Obtaining relevant set of trait specific genes from gene expression data is important for clinical diagnosis of disease and discovery of disease mechanisms in plants and animals. This process involves identification of relevant genes and removal of redundant genes as much as possible from a whole gene set. This package returns the trait specific gene set from the high dimensional RNA-seq count data by applying combination of two conventional machine learning algorithms, support vector machine (SVM) and genetic algorithm (GA). GA is used to control and optimize the subset of genes sent to the SVM for classification and evaluation. Genetic algorithm uses repeated learning steps and cross validation over number of possible solution and selects the best. The algorithm selects the set of genes based on a fitness function that is obtained via support vector machines. Using SVM as the classifier performance and the genetic algorithm for feature selection, a set of trait specific gene set is obtained.

This package provides a general framework for climate change vulnerability assessment (CCVA) of any species and any dataset. Variables are tagged as exposure, sensitivity or adaptive capacity; each component is computed from its own predictors by a chosen method; and the three components are combined into vulnerability. Methods range from the rule-based framework of Foden et al. (2013) <doi:10.1371/journal.pone.0065427> and the conventional weighted index of Ofori et al. (2017) <doi:10.1038/s41598-017-13245-y> to shape-constrained additive models ('scam'), latent-variable structural equation models ('lavaan') in which vulnerability is inferred from observable proxies, and penalised regression ('glmnet') for high-dimensional predictors. Missing traits and outcomes are handled by full-information maximum likelihood. Ecological knowledge enters as explicit constraints on predictors, on each component's role and on the links between components. Diagnostics, validation against observed outcomes, bootstrap and weight-based uncertainty, calibrated risk predictions and plots for every step are included.

Estimates synthetic control models while exposing the implementation choices that define the synthetic counterfactual: how matching variables are constructed, how predictor weights are set, which constraints the donor weights satisfy (simplex, lasso, ridge, penalized), whether an intercept is included, and which outcome model corrects for bias (OLS, ridge, lasso, augmented synthetic control). Runs specification-curve analysis over a grid of these choices (Simonsohn, Simmons and Nelson (2020) <doi:10.1038/s41562-020-0912-z>), with placebo-in-space and bootstrap inference for single specifications and placebo inference for the curve as a whole, and uses xgboost models with SHAP values (Lundberg and Lee (2017) <doi:10.48550/arXiv.1705.07874>) to show which choices drive the estimates. Methods follow Abadie, Diamond and Hainmueller (2010) <doi:10.1198/jasa.2009.ap08746>, Abadie (2021) <doi:10.1257/jel.20191450>, Ben-Michael, Feller and Rothstein (2021) <doi:10.1080/01621459.2021.1929245> and Chen and Guestrin (2016) <doi:10.1145/2939672.2939785>.

This package provides rank-based compositional differential abundance analysis for microbiome count data with zero cells using the compositional accelerated failure time model of Satten, Li and Zhao (2025) "CAFT: A Compositional Log-Linear Model for Microbiome Data with Zero Cells" <doi:10.1101/2025.11.26.690468>. Zero counts are treated as censored observations below sample-specific detection limits, avoiding the use of pseudocounts. The package implements estimation and hypothesis testing procedures for assessing associations between microbial taxa and covariates while accounting for the compositional structure of sequencing count data. It supports taxon-level differential abundance analysis, estimation of regression effects under censoring induced by detection limits, and inference based on rank-based methods that remain applicable in the presence of excess zeros. Functions are provided for model fitting, significance testing, extraction of effect estimates, and summarization of results across taxa. The package also provides optional bootstrap calibration of taxon-level p-values for sensitivity analysis in small-taxon settings.

Gives some hypothesis test functions (sign test, median and other quantile tests, Wilcoxon signed rank test, coefficient of variation test, test of normal variance, test on weighted sums of Poisson [see Fay and Kim <doi:10.1002/bimj.201600111>], sample size for t-tests with different variances and non-equal n per arm, Behrens-Fisher test, nonparametric ABC intervals, Wilcoxon-Mann-Whitney test [with effect estimates and confidence intervals, see Fay and Malinovsky <doi:10.1002/sim.7890>], two-sample melding tests [see Fay, Proschan, and Brittain <doi:10.1111/biom.12231>], one-way ANOVA allowing var.equal=FALSE [see Brown and Forsythe, 1974, Biometrics]), prevalence confidence intervals that adjust for sensitivity and specificity [see Lang and Reiczigel, 2014 <doi:10.1016/j.prevetmed.2013.09.015>] or Bayer, Fay, and Graubard, 2023 <doi:10.48550/arXiv.2205.13494>). The focus is on hypothesis tests that have compatible confidence intervals, but some functions only have confidence intervals (e.g., prevSeSp).

This package provides tools for interacting with the geographic name resolution service ('GNRS') API <https://github.com/ojalaquellueva/gnrs> and associated functionality. The GNRS is a batch application for resolving & standardizing political division names against standard name in the geonames database <http://www.geonames.org/>. The GNRS resolves political division names at three levels: country, state/province and county/parish. Resolution is performed in a series of steps, beginning with direct matching to standard names, followed by direct matching to alternate names in different languages, followed by direct matching to standard codes (such as ISO and FIPS codes). If direct matching fails, the GNRS attempts to match to standard and then alternate names using fuzzy matching, but does not perform fuzzing matching of political division codes. The GNRS works down the political division hierarchy, stopping at the current level if all matches fail. In other words, if a country cannot be matched, the GNRS does not attempt to match state or county.

Computes the prior fraction, the per-group pooling or shrinkage factor, for hierarchical models, including directly from brms fits. For each group-level coefficient the prior fraction is the share of the posterior precision contributed by the shrinkage prior relative to the likelihood; values near one indicate a coefficient that is prior-dominated (the centring/non-centring funnel regime), values near zero indicate a likelihood-dominated coefficient that is well identified from the data. These quantities are invisible to standard convergence diagnostics such as R-hat and effective sample size, and they indicate where a non-centred reparameterisation is likely to help. A companion advisor reports the same decomposition for changepoint random effects fitted with smoothbp'. The underlying geometry (the Fisher-metric connection on the base-fiber split, for which this connection is flat so the obstruction is statistical rather than geometric) is described in Bindoff (2026) <doi:10.5281/zenodo.20724550>; code reproducing the paper is in the package's source repository.

This package provides methods for estimating minimal important change (MIC) and interpretation thresholds for multi-item questionnaires and single-item continuous or ordinal measures. Methods include predictive modelling, adjusted predictive modelling, improved adjusted predictive modelling using anchor reliability, confirmatory factor analysis for anchor reliability, longitudinal confirmatory factor analysis for MIC estimation, longitudinal confirmatory factor analysis-based MIC estimation for single-item measures, and confirmatory factor analysis-based threshold estimation for single-item and multi-item measures. Implemented methods include those developed by Terluin et al. (2015) <doi:10.1016/j.jclinepi.2015.03.015>, Terluin et al. (2017) <doi:10.1016/j.jclinepi.2016.12.015>, Terluin et al. (2022) <doi:10.1016/j.jclinepi.2022.04.018>, Terluin et al. (2023) <doi:10.1007/s11136-023-03355-8>, Terluin et al. (2024) <doi:10.1007/s11136-023-03577-w>, Terluin et al. (2024) <doi:10.1007/s11136-024-03763-4>, and Terluin et al. (2026) <doi:10.1007/s11136-025-04134-3>.

This project aims to enable the method of Path Analysis to infer causalities from data. For this we propose a hybrid approach, which uses Bayesian network structure learning algorithms from data to create the input file for creation of a PA model. The process is performed in a semi-automatic way by our intermediate algorithm, allowing novice researchers to create and evaluate their own PA models from a data set. The references used for this project are: Koller, D., & Friedman, N. (2009). Probabilistic graphical models: principles and techniques. MIT press. <doi:10.1017/S0269888910000275>. Nagarajan, R., Scutari, M., & Lèbre, S. (2013). Bayesian networks in r. Springer, 122, 125-127. Scutari, M., & Denis, J. B. <doi:10.1007/978-1-4614-6446-4>. Scutari M (2010). Bayesian networks: with examples in R. Chapman and Hall/CRC. <doi:10.1201/b17065>. Rosseel, Y. (2012). lavaan: An R Package for Structural Equation Modeling. Journal of Statistical Software, 48(2), 1 - 36. <doi:10.18637/jss.v048.i02>.

Holistic Multimodel Domain Analysis (HMDA) is a robust and transparent framework designed for exploratory machine learning research, aiming to enhance the process of feature assessment and selection. HMDA addresses key limitations of traditional machine learning methods by evaluating the consistency across multiple high-performing models within a fine-tuned modeling grid, thereby improving the interpretability and reliability of feature importance assessments. Specifically, it computes Weighted Mean SHapley Additive exPlanations (WMSHAP), which aggregate feature contributions from multiple models based on weighted performance metrics. HMDA also provides confidence intervals to demonstrate the stability of these feature importance estimates. This framework is particularly beneficial for analyzing complex, multidimensional datasets common in health research, supporting reliable exploration of mental health outcomes such as suicidal ideation, suicide attempts, and other psychological conditions. Additionally, HMDA includes automated procedures for feature selection based on WMSHAP ratios and performs dimension reduction analyses to identify underlying structures among features. For more details see Haghish (2025) <doi:10.13140/RG.2.2.32473.63846>.

This package provides a set of models to estimate nonlinear longitudinal data using Bayesian estimation methods. These models include the: 1) Bayesian Piecewise Random Effects Model (Bayes_PREM()) which estimates a piecewise random effects (mixture) model for a given number of latent classes and a latent number of possible changepoints in each class, and can incorporate class and outcome predictive covariates (see Lamm (2022) <https://hdl.handle.net/11299/252533> and Lock et al., (2018) <doi:10.1007/s11336-017-9594-5>), 2) Bayesian Crossed Random Effects Model (Bayes_CREM()) which estimates a linear, quadratic, exponential, or piecewise crossed random effects models where individuals are changing groups over time (e.g., students and schools; see Rohloff et al., (2024) <doi:10.1111/bmsp.12334>), and 3) Bayesian Bivariate Piecewise Random Effects Model (Bayes_BPREM()) which estimates a bivariate piecewise random effects model to jointly model two related outcomes (e.g., reading and math achievement; see Peralta et al., (2022) <doi:10.1037/met0000358>).

Specify, solve, and estimate dynamic stochastic general equilibrium (DSGE) models by maximum likelihood and Bayesian methods. Supports both linear models via an equation-based formula interface and nonlinear models via string-based equations with perturbation up to third order (Schmitt-Grohe and Uribe, 2004 <doi:10.1016/S0165-1889(03)00043-5>). Solution uses the method of undetermined coefficients (Klein, 2000 <doi:10.1016/S0165-1889(99)00045-7>). Likelihood evaluated via the Kalman filter or a bootstrap particle filter (Gordon et al., 1993). Bayesian estimation uses adaptive Random-Walk Metropolis-Hastings or Particle Marginal Metropolis-Hastings (Andrieu et al., 2010 <doi:10.1111/j.1467-9868.2009.00736.x>) with parallel chain support. Additional tools include Bayes factor model comparison with Kass-Raftery evidence scales, Ramsey optimal policy via linear-quadratic regulator, nonlinear perfect foresight via stacked-time Newton (Juillard et al., 1998), Kalman smoothing, historical shock decomposition, local identification diagnostics, parameter sensitivity analysis, occasionally binding constraints, impulse-response functions, forecasting, and robust standard errors.

This package provides nonparametric CUSUM tests for detecting changes in possibly serially dependent univariate or low-dimensional multivariate observations. Retrospective tests sensitive to changes in the expectation, the variance, the covariance, the autocovariance, the distribution function, Spearman's rho, Kendall's tau, Gini's mean difference, and the copula are provided, as well as a test for detecting changes in the distribution of independent block maxima (with environmental studies in mind). The package also contains a test sensitive to changes in the autocopula and a combined test of stationarity sensitive to changes in the distribution function and the autocopula. The latest additions are an open-end sequential test based on the retrospective CUSUM statistic that can be used for monitoring changes in the mean of possibly serially dependent univariate observations, as well as closed-end and open-end sequential tests based on empirical distribution functions that can be used for monitoring changes in the contemporary distribution of possibly serially dependent univariate or low-dimensional multivariate observations.

Ordination comprises several multivariate exploratory and explanatory techniques with theoretical foundations in geometric data analysis; see Podani (2000, ISBN:90-5782-067-6) for techniques and applications and Le Roux & Rouanet (2005) <doi:10.1007/1-4020-2236-0> for foundations. Greenacre (2010, ISBN:978-84-923846) shows how the most established of these, including principal components analysis, correspondence analysis, multidimensional scaling, factor analysis, and discriminant analysis, rely on eigen-decompositions or singular value decompositions of pre-processed numeric matrix data. These decompositions give rise to a set of shared coordinates along which the row and column elements can be measured. The overlay of their scatterplots on these axes, introduced by Gabriel (1971) <doi:10.1093/biomet/58.3.453>, is called a biplot. ordr provides inspection, extraction, manipulation, and visualization tools for several popular ordination classes supported by a set of recovery methods. It is inspired by and designed to integrate into Tidyverse workflows provided by Wickham et al (2019) <doi:10.21105/joss.01686>.

This package provides tools for modelling and forecasting epidemic trajectories using a dynamic Gompertz model within a state space framework, with the Kalman filter for robust estimation of non-linear growth. Includes a reinitialization feature to adapt to new waves, and a leading-indicator extension that uses a related series moving ahead of the variable of interest (e.g. cases ahead of hospitalisations) to improve short-horizon forecasts, with model and lag selection via rolling-origin cross-validation. Applicable to data at daily, monthly, quarterly, or annual frequency, and to non-epidemic trajectories with similar dynamics, such as innovation diffusion and product adoption. Includes functions for data preprocessing, model fitting, forecast visualization, and accuracy evaluation using standard error measures. Methods are described in Harvey and Kattuman (2020) <doi:10.1162/99608f92.828f40de>, Harvey and Kattuman (2021) <doi:10.1098/rsif.2021.0179>, and Ashby, Harvey, Kattuman, Tang, and Thamotheram (2024) <https://www.jbs.cam.ac.uk/wp-content/uploads/2024/03/cchle-tsgc-paper-2024.pdf>.

Tests whether a condition is both necessary and sufficient for an outcome, that is, whether the outcome cannot be reached without the condition and is guaranteed in its presence. Such a claim implies that two diagonally opposite corners of the condition-outcome scatter plot contain no observations. The package estimates both boundaries with the same data, scope and frontier technique, reports the effect size and permutation test of each, and indicates joint support only when both components are supported. The two components are summarized by the weaker effect, a balanced combination and the proportion of the scope that the two claims jointly rule out. A dual-threshold table gives, for each outcome level, the condition level it requires, the level sufficient for it, and the interval between them. Results can be compared with regression slopes. The package also provides plots, influential-case diagnostics, scope-sensitivity and resampling checks, and simulation and power analysis for planning studies. The two sides are estimated with the NCA and SCAtools packages.

This package implements Moreau-Yosida Markov chain Monte Carlo ('MCMC') importance sampling for parameter estimation and Bayesian inference under smooth, non-differentiable, or light-tailed target posterior distributions and arbitrary probability models with complete or censored data. Users supply user-defined probability density functions, optional distribution functions, parameter ranges, and observations subject to complete, right, left, interval, Type-I, Type-II, progressive Type-II, first-failure, or truncation schemes. Constructs Moreau-Yosida envelopes, gradient-based proposals ('MALA', HMC', or RWM'), self-normalized importance weights, batch-means asymptotic variance estimates, and Bayesian marginal quantiles. Methodologies are based on Shukla', Vats', and Chi (2025) <doi:10.48550/arXiv.2501.02228>, Pereyra (2016) <doi:10.1111/sjos.12208>, Durmus and others (2022) <doi:10.1214/22-EJS2027>, Chen and Shao (1999) <doi:10.1214/ss/1009211804>, Roberts and Rosenthal (1998) <doi:10.1214/aoap/1028903378>, Geweke (1989) <doi:10.2307/2290062>, Hesterberg (1995) <doi:10.1080/00031305.1995.10476138>, and Balakrishnan and Aggarwala (2000, ISBN:978-0-8176-4001-9).

Implementations of an estimator for the multivariate regression association measure (MRAM) proposed in Shih and Chen (2026) <doi:10.1016/j.csda.2025.108288> and its associated variable selection algorithm. The MRAM quantifies the predictability of a random vector Y from a random vector X given a random vector Z. It takes the maximum value 1 if and only if Y is almost surely a measurable function of X and Z, and the minimum value of 0 if Y is conditionally independent of X given Z. The MRAM generalizes the Kendall's tau copula correlation ratio proposed in Shih and Emura (2021) <doi:10.1016/j.jmva.2020.104708> by employing the spatial sign function. The estimator is based on the nearest neighbor method, and the associated variable selection algorithm is adapted from the feature ordering by conditional independence (FOCI) algorithm of Azadkia and Chatterjee (2021) <doi:10.1214/21-AOS2073>. For further details, see the paper Shih and Chen (2026) <doi:10.1016/j.csda.2025.108288>.