This package provides a framework for specifying and running flexible linear-time reachability-based algorithms for graphical causal inference. Rule tables are used to encode and customize the reachability algorithm to typical causal and probabilistic reasoning tasks such as finding d-connected nodes or more advanced applications. For more information, see Wienöbst, Weichwald and Henckel (2025) <doi:10.48550/arXiv.2506.15758>.

Interface to the Google Cloud Machine Learning Platform <https://cloud.google.com/vertex-ai>, which provides cloud tools for training machine learning models.

An implementation of methods for causal discovery in a structural causal model where the conditional distribution of the target node is described by a generalized linear model conditional on its causal parents.

Conditioned Latin hypercube sampling, as published by Minasny and McBratney (2006) <DOI:10.1016/j.cageo.2005.12.009>. This method proposes to stratify sampling in presence of ancillary data. An extension of this method, which propose to associate a cost to each individual and take it into account during the optimisation process, is also proposed (Roudier et al., 2012, <DOI:10.1201/b12728>).

Utilities that support the usage of pyDarwin (<https://certara.github.io/pyDarwin/>) for ease of setup and execution of a machine learning based pharmacometric model search with Certara's Non-Linear Mixed Effects (NLME) modeling engine.

This package provides functions for performing experimental comparisons of algorithms using adequate sample sizes for power and accuracy. Implements the methodology originally presented in Campelo and Takahashi (2019) <doi:10.1007/s10732-018-9396-7> for the comparison of two algorithms, and later generalised in Campelo and Wanner (Submitted, 2019) <arxiv:1908.01720>.

This package provides a suite of computer model test functions that can be used to test and evaluate algorithms for Bayesian (also known as sequential) optimization. Some of the functions have known functional forms, however, most are intended to serve as black-box functions where evaluation requires running computer code that reveals little about the functional forms of the objective and/or constraints. The primary goal of the package is to provide users (especially those who do not have access to real computer models) a source of reproducible and shareable examples that can be used for benchmarking algorithms. The package is a living repository, and so more functions will be added over time. For function suggestions, please do contact the author of the package.

Implementation of the categorical instrumental variable (CIV) estimator proposed by Wiemann (2023) <arXiv:2311.17021>. CIV allows for optimal instrumental variable estimation in settings with relatively few observations per category. To obtain valid inference in these challenging settings, CIV leverages a regularization assumption that implies existence of a latent categorical variable with fixed finite support achieving the same first stage fit as the observed instrument.

This package implements a Bayesian approach to causal impact estimation in time series, as described in Brodersen et al. (2015) <DOI:10.1214/14-AOAS788>. See the package documentation on GitHub <https://google.github.io/CausalImpact/> to get started.

The Codemeta Project defines a JSON-LD format for describing software metadata, as detailed at <https://codemeta.github.io>. This package provides core utilities to generate this metadata with a minimum of dependencies.

Takes the outputs of a caret confusion matrix and allows for the quick conversion of these list items to lists. The intended usage is to allow the tool to work with the outputs of machine learning classification models. This tool works with classification problems for binary and multi-classification problems and allows for the record level conversion of the confusion matrix outputs. This is useful, as it allows quick conversion of these objects for storage in database systems and to track ML model performance over time. Traditionally, this approach has been used for highlighting model representation and feature slippage.

This package provides functions calculating Conley (1999) <doi:10.1016/S0304-4076(98)00084-0> standard errors. The package started by merging and extending multiple packages and other published scripts on this econometric technique. It strongly emphasizes computational optimization. Details are available in the function documentation and in the vignette.

This package performs adjustments of a user-supplied independence loglikelihood function using a robust sandwich estimator of the parameter covariance matrix, based on the methodology in Chandler and Bate (2007) <doi:10.1093/biomet/asm015>. This can be used for cluster correlated data when interest lies in the parameters of the marginal distributions or for performing inferences that are robust to certain types of model misspecification. Functions for profiling the adjusted loglikelihoods are also provided, as are functions for calculating and plotting confidence intervals, for single model parameters, and confidence regions, for pairs of model parameters. Nested models can be compared using an adjusted likelihood ratio test.

In statistical modeling, multiple models need to be compared based on certain criteria. The method described here uses eight metrics from AllMetrics package. â input_dfâ is the data frame (at least two columns for comparison) containing metrics values in different rows of a column (which denotes a particular modelâ s performance). First five metrics are expected to be minimum and last three metrics are expected to be maximum for a model to be considered good. Firstly, every metric value (among first five) is searched in every columns and minimum values are denoted as â MINâ and other values are denoted as â NAâ . Secondly, every metric (among last three) is searched in every columns and maximum values are denoted as â MAXâ and other values are denoted as â NAâ . â output_dfâ contains the similar number of rows (which is 8) and columns (which is number of models to be compared) as of â input_dfâ . Values in â output_dfâ are corresponding â NAâ , â MINâ or â MAXâ . Finally, the column containing minimum number of â NAâ values is denoted as the best column. â min_NA_colâ gives the name of the best column (model). â min_NA_valuesâ are the corresponding metrics values. âBestColumn_metricsâ is the data frame (dimension: 1*8) containing different metrics of the best column (model). â best_column_resultsâ is the final result (a list) containing all of these output elements. In special case, if two columns having equal NA', it will be checked among these two column which one is having least NA in first five rows and will be inferred as the best. More details about AllMetrics can be found in Garai (2023) <doi:10.13140/RG.2.2.18688.30723>.

For Bayesian and classical inference and prediction with count-valued data, Simultaneous Transformation and Rounding (STAR) Models provide a flexible, interpretable, and easy-to-use approach. STAR models the observed count data using a rounded continuous data model and incorporates a transformation for greater flexibility. Implicitly, STAR formalizes the commonly-applied yet incoherent procedure of (i) transforming count-valued data and subsequently (ii) modeling the transformed data using Gaussian models. STAR is well-defined for count-valued data, which is reflected in predictive accuracy, and is designed to account for zero-inflation, bounded or censored data, and over- or underdispersion. Importantly, STAR is easy to combine with existing MCMC or point estimation methods for continuous data, which allows seamless adaptation of continuous data models (such as linear regressions, additive models, BART, random forests, and gradient boosting machines) for count-valued data. The package also includes several methods for modeling count time series data, namely via warped Dynamic Linear Models. For more details and background on these methodologies, see the works of Kowal and Canale (2020) <doi:10.1214/20-EJS1707>, Kowal and Wu (2022) <doi:10.1111/biom.13617>, King and Kowal (2022) <arXiv:2110.14790>, and Kowal and Wu (2023) <arXiv:2110.12316>.

Gene Symbols or Ensembl Gene IDs are converted using the Bimap interface in AnnotationDbi in convertId2() but that function is only provided as fallback mechanism for the most common use cases in data analysis. The main function in the package is convert.bm() which queries BioMart using the full capacity of the API provided through the biomaRt package. Presets and defaults are provided for convenience but all "marts", "filters" and "attributes" can be set by the user. Function convert.alias() converts Gene Symbols to Aliases and vice versa and function likely_symbol() attempts to determine the most likely current Gene Symbol.

This package contains functions to detect and visualise periods of climate sensitivity (climate windows) for a given biological response. Please see van de Pol et al. (2016) <doi:10.1111/2041-210X.12590> and Bailey and van de Pol (2016) <doi:10.1371/journal.pone.0167980> for details.

This package provides a wrapper for the CDRC API that returns data frames or sf of CDRC data. The API web reference is:<https://api.cdrc.ac.uk/swagger/index.html>.

Variance estimation on indicators of income concentration and poverty using complex sample survey designs. Wrapper around the survey package.

This package provides a specialized tool is designed for assessing contextual bandit algorithms, particularly those aimed at handling overdispersed and zero-inflated count data. It offers a simulated testing environment that includes various models like Poisson, Overdispersed Poisson, Zero-inflated Poisson, and Zero-inflated Overdispersed Poisson. The package is capable of executing five specific algorithms: Linear Thompson sampling with log transformation on the outcome, Thompson sampling Poisson, Thompson sampling Negative Binomial, Thompson sampling Zero-inflated Poisson, and Thompson sampling Zero-inflated Negative Binomial. Additionally, it can generate regret plots to evaluate the performance of contextual bandit algorithms. This package is based on the algorithms by Liu et al. (2023) <arXiv:2311.14359>.

Inference with control function methods for nonlinear outcome models when the model is known ('Guo and Small (2016) <arXiv:1602.01051>) and when unknown but semiparametric ('Li and Guo (2021) <arXiv:2010.09922>).

Customized training is a simple technique for transductive learning, when the test covariates are known at the time of training. The method identifies a subset of the training set to serve as the training set for each of a few identified subsets in the training set. This package implements customized training for the glmnet() and cv.glmnet() functions.

This package provides functions for cobin and micobin regression models, a new family of generalized linear models for continuous proportional data (Y in the closed unit interval [0, 1]). It also includes an exact, efficient sampler for the Kolmogorov-Gamma random variable. For details, see Lee et al. (2025+) <doi:10.48550/arXiv.2504.15269>.

Evaluate arbitrary function calls using workers on HPC schedulers in single line of code. All processing is done on the network without accessing the file system. Remote schedulers are supported via SSH.