Prepares data for statistical analysis (e.g., analysis of variance ;ANOVA) by enabling the user to easily and quickly merge (using the file_merge() function) raw data files into one merged table and then aggregate the merged table (using the prep() function) into a finalized table while keeping track and summarizing every step of the preparation. The finalized table contains several possibilities for dependent measures of the dependent variable. Most suitable when measuring variables in an interval or ratio scale (e.g., reaction-times) and/or discrete values such as accuracy. Main functions included are file_merge() and prep(). The file_merge() function vertically merges individual data files (in a long format) in which each line is a single observation to one single dataset. The prep() function aggregates the single dataset according to any combination of grouping variables (i.e., between-subjects and within-subjects independent variables, respectively), and returns a data frame with a number of dependent measures for further analysis for each cell according to the combination of provided grouping variables. Dependent measures for each cell include among others means before and after rejecting all values according to a flexible standard deviation criteria, number of rejected values according to the flexible standard deviation criteria, proportions of rejected values according to the flexible standard deviation criteria, number of values before rejection, means after rejecting values according to procedures described in Van Selst & Jolicoeur (1994; suitable when measuring reaction-times), standard deviations, medians, means according to any percentile (e.g., 0.05, 0.25, 0.75, 0.95) and harmonic means. The data frame prep() returns can also be exported as a txt file to be used for statistical analysis in other statistical programs.

Cross validation informed Relaxed LASSO (or more generally elastic net), gradient boosting machine ('xgboost'), Random Forest ('RandomForestSRC'), Oblique Random Forest ('aorsf'), Artificial Neural Network (ANN), Recursive Partitioning ('RPART') or step wise regression models are fit. Cross validation leave out samples (leading to nested cross validation) or bootstrap out-of-bag samples are used to evaluate and compare performances between these models with results presented in tabular or graphical means. Calibration plots can also be generated, again based upon (outer nested) cross validation or bootstrap leave out (out of bag) samples. Note, at the time of this writing, in order to fit gradient boosting machine models one must install the packages DiceKriging and rgenoud using the install.packages() function. For some datasets, for example when the design matrix is not of full rank, glmnet may have very long run times when fitting the relaxed lasso model, from our experience when fitting Cox models on data with many predictors and many patients, making it difficult to get solutions from either glmnet() or cv.glmnet(). This may be remedied by using the path=TRUE option when calling glmnet() and cv.glmnet(). Within the glmnetr package the approach of path=TRUE is taken by default. other packages doing similar include nestedcv <https://cran.r-project.org/package=nestedcv>, glmnetSE <https://cran.r-project.org/package=glmnetSE> which may provide different functionality when performing a nested CV. Use of the glmnetr has many similarities to the glmnet package and it could be helpful for the user of glmnetr also become familiar with the glmnet package <https://cran.r-project.org/package=glmnet>, with the "An Introduction to glmnet'" and "The Relaxed Lasso" being especially useful in this regard.

Draws forest plots of exposure effects from fitted regression models. Name an exposure and foresty plots its effect. Name an effect modifier as well and it refits the model with the interaction term, estimates the exposure effect within each level of the modifier as a linear combination of the coefficients, and reports the joint interaction test beside those estimates. It takes one exposure and one modifier at a time, so the interaction is always a two-way one. Rows of the plot and of the table beside it share one scale, in a layout that can follow a journal's house style. The same results go to a self-contained HTML page holding the subgroup estimates, the joint test and the coefficient table. The car package computes the linear combinations and their tests. Models fitted by stats::glm(), stats::lm(), the survival package, the lme4 package, the geepack package and the survey package are supported, as is any fit supplying coef() and vcov(). Ordinal outcomes are supported through the MASS package and nominal ones through the nnet package, where the figure carries one row per level of the outcome and the interaction is tested jointly across the equations. Fits from the rms package are refused, naming the function that fits the same model in their place. The estimation of an exposure effect within a level of a modifier, and the test of the difference between such estimates, follow Altman and Bland (2003) <doi:10.1136/bmj.326.7382.219> and VanderWeele and Knol (2014) <doi:10.1515/em-2013-0005>; the reporting of subgroup effects beside the interaction test follows Wang et al. (2007) <doi:10.1056/NEJMsr077003>, and the figure itself the forest plot described by Lewis and Clarke (2001) <doi:10.1136/bmj.322.7300.1479>.

Calculate multiple biotic indices using diatoms from environmental samples. Diatom species are recognized by their species name using a heuristic search, and their ecological data is retrieved from multiple sources. It includes number/shape of chloroplasts diversity indices, size classes, ecological guilds, and multiple biotic indices. It outputs both a dataframe with all the results and plots of all the obtained data in a defined output folder. - Sample data was taken from Nicolosi Gelis, Cochero & Gómez (2020, <doi:10.1016/j.ecolind.2019.105951>). - The package uses the Diat.Barcode database to calculate morphological and ecological information by Rimet & Couchez (2012, <doi:10.1051/kmae/2012018>),and the combined classification of guilds and size classes established by B-Béres et al. (2017, <doi:10.1016/j.ecolind.2017.07.007>). - Current diatom-based biotic indices include the DES index by Descy (1979) - EPID index by Dell'Uomo (1996, ISBN: 3950009002) - IDAP index by Prygiel & Coste (1993, <doi:10.1007/BF00028033>) - ID-CH index by Hürlimann & Niederhauser (2007) - IDP index by Gómez & Licursi (2001, <doi:10.1023/A:1011415209445>) - ILM index by Leclercq & Maquet (1987) - IPS index by Coste (1982) - LOBO index by Lobo, Callegaro, & Bender (2002, ISBN:9788585869908) - SLA by SládeÄ ek (1986, <doi:10.1002/aheh.19860140519>) - TDI index by Kelly, & Whitton (1995, <doi:10.1007/BF00003802>) - SPEAR(herbicide) index by Wood, Mitrovic, Lim, Warne, Dunlop, & Kefford (2019, <doi:10.1016/j.ecolind.2018.12.035>) - PBIDW index by Castro-Roa & Pinilla-Agudelo (2014) - DISP index by Stenger-Kovács et al. (2018, <doi:10.1016/j.ecolind.2018.07.026>) - EDI index by Chamorro et al. (2024, <doi:10.1021/acsestwater.4c00126>) - DDI index by à lvarez-Blanco et al. (2013, <doi: 10.1007/s10661-012-2607-z>) - PDISE index by Kahlert et al. (2023, <doi:10.1007/s10661-023-11378-4>).

Analysis of task-related functional magnetic resonance imaging (fMRI) activity at the level of individual participants is commonly based on general linear modelling (GLM) that allows us to estimate to what extent the blood oxygenation level dependent (BOLD) signal can be explained by task response predictors specified in the GLM model. The predictors are constructed by convolving the hypothesised timecourse of neural activity with an assumed hemodynamic response function (HRF). To get valid and precise estimates of task response, it is important to construct a model of neural activity that best matches actual neuronal activity. The construction of models is most often driven by predefined assumptions on the components of brain activity and their duration based on the task design and specific aims of the study. However, our assumptions about the onset and duration of component processes might be wrong and can also differ across brain regions. This can result in inappropriate or suboptimal models, bad fitting of the model to the actual data and invalid estimations of brain activity. Here we present an approach in which theoretically driven models of task response are used to define constraints based on which the final model is derived computationally using the actual data. Specifically, we developed autohrf â a package for the R programming language that allows for data-driven estimation of HRF models. The package uses genetic algorithms to efficiently search for models that fit the underlying data well. The package uses automated parameter search to find the onset and duration of task predictors which result in the highest fitness of the resulting GLM based on the fMRI signal under predefined restrictions. We evaluate the usefulness of the autohrf package on publicly available datasets of task-related fMRI activity. Our results suggest that by using autohrf users can find better task related brain activity models in a quick and efficient manner.

Simulation methods for phylogenetic trees where (i) all tips are sampled at one time point or (ii) tips are sampled sequentially through time. (i) For sampling at one time point, simulations are performed under a constant rate birth-death process, conditioned on having a fixed number of final tips (sim.bd.taxa()), or a fixed age (sim.bd.age()), or a fixed age and number of tips (sim.bd.taxa.age()). When conditioning on the number of final tips, the method allows for shifts in rates and mass extinction events during the birth-death process (sim.rateshift.taxa()). The function sim.bd.age() (and sim.rateshift.taxa() without extinction) allow the speciation rate to change in a density-dependent way. The LTT plots of the simulations can be displayed using LTT.plot(), LTT.plot.gen() and LTT.average.root(). TreeSim further samples trees with n final tips from a set of trees generated by the common sampling algorithm stopping when a fixed number m>>n of tips is first reached (sim.gsa.taxa()). This latter method is appropriate for m-tip trees generated under a big class of models (details in the sim.gsa.taxa() man page). For incomplete phylogeny, the missing speciation events can be added through simulations (corsim()). (ii) sim.rateshifts.taxa() is generalized to sim.bdsky.stt() for serially sampled trees, where the trees are conditioned on either the number of sampled tips or the age. Furthermore, for a multitype-branching process with sequential sampling, trees on a fixed number of tips can be simulated using sim.bdtypes.stt.taxa(). This function further allows to simulate under epidemiological models with an exposed class. The function sim.genespeciestree() simulates coalescent gene trees within birth-death species trees, and sim.genetree() simulates coalescent gene trees.

Measures morphological diversity from discrete character data and estimates evolutionary tempo on phylogenetic trees. Imports morphological data from #NEXUS (Maddison et al. (1997) <doi:10.1093/sysbio/46.4.590>) format with read_nexus_matrix(), and writes to both #NEXUS and TNT format (Goloboff et al. (2008) <doi:10.1111/j.1096-0031.2008.00217.x>). Main functions are test_rates(), which implements AIC and likelihood ratio tests for discrete character rates introduced across Lloyd et al. (2012) <doi:10.1111/j.1558-5646.2011.01460.x>, Brusatte et al. (2014) <doi:10.1016/j.cub.2014.08.034>, Close et al. (2015) <doi:10.1016/j.cub.2015.06.047>, and Lloyd (2016) <doi:10.1111/bij.12746>, and calculate_morphological_distances(), which implements multiple discrete character distance metrics from Gower (1971) <doi:10.2307/2528823>, Wills (1998) <doi:10.1006/bijl.1998.0255>, Lloyd (2016) <doi:10.1111/bij.12746>, and Hopkins and St John (2018) <doi:10.1098/rspb.2018.1784>. This also includes the GED correction from Lehmann et al. (2019) <doi:10.1111/pala.12430>. Multiple functions implement morphospace plots: plot_chronophylomorphospace() implements Sakamoto and Ruta (2012) <doi:10.1371/journal.pone.0039752>, plot_morphospace() implements Wills et al. (1994) <doi:10.1017/S009483730001263X>, plot_changes_on_tree() implements Wang and Lloyd (2016) <doi:10.1098/rspb.2016.0214>, and plot_morphospace_stack() implements Foote (1993) <doi:10.1017/S0094837300015864>. Other functions include safe_taxonomic_reduction(), which implements Wilkinson (1995) <doi:10.1093/sysbio/44.4.501>, map_dollo_changes() implements the Dollo stochastic character mapping of Tarver et al. (2018) <doi:10.1093/gbe/evy096>, and estimate_ancestral_states() implements the ancestral state options of Lloyd (2018) <doi:10.1111/pala.12380>. calculate_tree_length() and reconstruct_ancestral_states() implements the generalised algorithms from Swofford and Maddison (1992; no doi).

Fast partial least squares (PLS) for dense and out-of-core data. Provides SIMPLS (straightforward implementation of a statistically inspired modification of the PLS method) and NIPALS (non-linear iterative partial least-squares) solvers, plus kernel-style PLS variants ('kernelpls and widekernelpls') with parity to pls'. Optimized for bigmemory'-backed matrices with streamed cross-products and chunked BLAS (Basic Linear Algebra Subprograms) (XtX/XtY and XXt/YX), optional file-backed score sinks, and deterministic testing helpers. Includes an auto-selection strategy that chooses between XtX SIMPLS, XXt (wide) SIMPLS, and NIPALS based on (n, p) and a configurable memory budget. About the package, Bertrand and Maumy (2023) <https://hal.science/hal-05352069>, and <https://hal.science/hal-05352061> highlighted fitting and cross-validating PLS regression models to big data. For more details about some of the techniques featured in the package, Dayal and MacGregor (1997) <doi:10.1002/(SICI)1099-128X(199701)11:1%3C73::AID-CEM435%3E3.0.CO;2-%23>, Rosipal & Trejo (2001) <https://www.jmlr.org/papers/v2/rosipal01a.html>, Tenenhaus, Viennet, and Saporta (2007) <doi:10.1016/j.csda.2007.01.004>, Rosipal (2004) <doi:10.1007/978-3-540-45167-9_17>, Song, Wang, and Bai (2024) <doi:10.1016/j.chemolab.2024.105238>. Includes kernel logistic PLS with C++'-accelerated alternating iteratively reweighted least squares (IRLS) updates, streamed reproducing kernel Hilbert space (RKHS) solvers with reusable centering statistics, and bootstrap diagnostics with graphical summaries for coefficients, scores, and cross-validation workflows, alongside dedicated plotting utilities for individuals, variables, ellipses, and biplots. The streaming backend uses far less memory and keeps memory bounded across data sizes. For PLS1, streaming is often fast enough while preserving a small memory footprint; for PLS2 it remains competitive with a bounded footprint. On small problems that fit comfortably in RAM (random-access memory), dense in-memory solvers are slightly faster; the crossover occurs as n or p grow and the Gram/cross-product cost dominates.

This package provides a robust implementation of Topolow algorithm. It embeds objects into a low-dimensional Euclidean space from a matrix of pairwise dissimilarities, even when the data do not satisfy metric or Euclidean axioms. The package is particularly well-suited for sparse, incomplete, and censored (thresholded) datasets such as antigenic relationships. The core is a physics-inspired, gradient-free optimization framework that models objects as particles in a physical system, where observed dissimilarities define spring rest lengths and unobserved pairs exert repulsive forces. The package also provides functions specific to antigenic mapping to transform cross-reactivity and binding affinity measurements into accurate spatial representations in a phenotype space. Key features include: * Robust Embedding from Sparse Data: Effectively creates complete and consistent maps (in optimal dimensions) even with high proportions of missing data (e.g., >95%). * Physics-Inspired Optimization: Models objects (e.g., antigens, landmarks) as particles connected by springs (for measured dissimilarities) and subject to repulsive forces (for missing dissimilarities), and simulates the physical system using laws of mechanics, reducing the need for complex gradient computations. * Automatic Dimensionality Detection: Employs a likelihood-based approach to determine the optimal number of dimensions for the embedding/map, avoiding distortions common in methods with fixed low dimensions. * Noise and Bias Reduction: Naturally mitigates experimental noise and bias through its network-based, error-dampening mechanism. * Antigenic Velocity Calculation (for antigenic data): Introduces and quantifies "antigenic velocity," a vector that describes the rate and direction of antigenic drift for each pathogen isolate. This can help identify cluster transitions and potential lineage replacements. * Broad Applicability: Analyzes data from various objects that their dissimilarity may be of interest, ranging from complex biological measurements such as continuous and relational phenotypes, antibody-antigen interactions, and protein folding to abstract concepts, such as customer perception of different brands. Methods are described in the context of bioinformatics applications in Arhami and Rohani (2025a) <doi:10.1093/bioinformatics/btaf372>, and mathematical proofs and Euclidean embedding details are in Arhami and Rohani (2025b) <doi:10.48550/arXiv.2508.01733>.

Analyzing the performance of artificial intelligence (AI) systems/algorithms characterized by a search-and-report strategy. Historically observer performance has dealt with measuring radiologists performances in search tasks, e.g., searching for lesions in medical images and reporting them, but the implicit location information has been ignored. The implemented methods apply to analyzing the absolute and relative performances of AI systems, comparing AI performance to a group of human readers or optimizing the reporting threshold of an AI system. In addition to performing historical receiver operating receiver operating characteristic (ROC) analysis (localization information ignored), the software also performs free-response receiver operating characteristic (FROC) analysis, where lesion localization information is used. A book using the software has been published: Chakraborty DP: Observer Performance Methods for Diagnostic Imaging - Foundations, Modeling, and Applications with R-Based Examples, Taylor-Francis LLC; 2017: <https://www.routledge.com/Observer-Performance-Methods-for-Diagnostic-Imaging-Foundations-Modeling/Chakraborty/p/book/9781482214840>. Online updates to this book, which use the software, are at <https://dpc10ster.github.io/RJafrocQuickStart/>, <https://dpc10ster.github.io/RJafrocRocBook/> and at <https://dpc10ster.github.io/RJafrocFrocBook/>. Supported data collection paradigms are the ROC, FROC and the location ROC (LROC). ROC data consists of single ratings per images, where a rating is the perceived confidence level that the image is that of a diseased patient. An ROC curve is a plot of true positive fraction vs. false positive fraction. FROC data consists of a variable number (zero or more) of mark-rating pairs per image, where a mark is the location of a reported suspicious region and the rating is the confidence level that it is a real lesion. LROC data consists of a rating and a location of the most suspicious region, for every image. Four models of observer performance, and curve-fitting software, are implemented: the binormal model (BM), the contaminated binormal model (CBM), the correlated contaminated binormal model (CORCBM), and the radiological search model (RSM). Unlike the binormal model, CBM, CORCBM and RSM predict proper ROC curves that do not inappropriately cross the chance diagonal. Additionally, RSM parameters are related to search performance (not measured in conventional ROC analysis) and classification performance. Search performance refers to finding lesions, i.e., true positives, while simultaneously not finding false positive locations. Classification performance measures the ability to distinguish between true and false positive locations. Knowing these separate performances allows principled optimization of reader or AI system performance. This package supersedes Windows JAFROC (jackknife alternative FROC) software V4.2.1, <https://github.com/dpc10ster/WindowsJafroc>. Package functions are organized as follows. Data file related function names are preceded by Df', curve fitting functions by Fit', included data sets by dataset', plotting functions by Plot', significance testing functions by St', sample size related functions by Ss', data simulation functions by Simulate and utility functions by Util'. Implemented are figures of merit (FOMs) for quantifying performance and functions for visualizing empirical or fitted operating characteristics: e.g., ROC, FROC, alternative FROC (AFROC) and weighted AFROC (wAFROC) curves. For fully crossed study designs significance testing of reader-averaged FOM differences between modalities is implemented via either Dorfman-Berbaum-Metz or the Obuchowski-Rockette methods. Also implemented is single treatment analysis, which allows comparison of performance of a group of radiologists to a specified value, or comparison of AI to a group of radiologists interpreting the same cases. Crossed-modality analysis is implemented wherein there are two crossed treatment factors and the aim is to determined performance in each treatment factor averaged over all levels of the second factor. Sample size estimation tools are provided for ROC and FROC studies; these use estimates of the relevant variances from a pilot study to predict required numbers of readers and cases in a pivotal study to achieve the desired power. Utility and data file manipulation functions allow data to be read in any of the currently used input formats, including Excel, and the results of the analysis can be viewed in text or Excel output files. The methods are illustrated with several included datasets from the author's collaborations. This update includes improvements to the code, some as a result of user-reported bugs and new feature requests, and others discovered during ongoing testing and code simplification.

Flexible multidimensional scaling (MDS) methods and extensions to the package smacof'. This package contains various functions, wrappers, methods and classes for fitting, plotting and displaying a large number of different flexible MDS models. These are: Torgerson scaling (Torgerson, 1958, ISBN:978-0471879459) with powers, Sammon mapping (Sammon, 1969, <doi:10.1109/T-C.1969.222678>) with ratio and interval optimal scaling, Multiscale MDS (Ramsay, 1977, <doi:10.1007/BF02294052>) with ratio and interval optimal scaling, s-stress MDS (ALSCAL; Takane, Young & De Leeuw, 1977, <doi:10.1007/BF02293745>) with ratio and interval optimal scaling, elastic scaling (McGee, 1966, <doi:10.1111/j.2044-8317.1966.tb00367.x>) with ratio and interval optimal scaling, r-stress MDS (De Leeuw, Groenen & Mair, 2016, <https://rpubs.com/deleeuw/142619>) with ratio, interval, splines and nonmetric optimal scaling, power-stress MDS (POST-MDS; Buja & Swayne, 2002 <doi:10.1007/s00357-001-0031-0>) with ratio and interval optimal scaling, restricted power-stress (Rusch, Mair & Hornik, 2021, <doi:10.1080/10618600.2020.1869027>) with ratio and interval optimal scaling, approximate power-stress with ratio optimal scaling (Rusch, Mair & Hornik, 2021, <doi:10.1080/10618600.2020.1869027>), Box-Cox MDS (Chen & Buja, 2013, <https://jmlr.org/papers/v14/chen13a.html>), local MDS (Chen & Buja, 2009, <doi:10.1198/jasa.2009.0111>), curvilinear component analysis (Demartines & Herault, 1997, <doi:10.1109/72.554199>), curvilinear distance analysis (Lee, Lendasse & Verleysen, 2004, <doi:10.1016/j.neucom.2004.01.007>), nonlinear MDS with optimal dissimilarity powers functions (De Leeuw, 2024, <https://github.com/deleeuw/smacofManual/blob/main/smacofPO(power)/smacofPO.pdf>), sparsified (power) MDS and sparsified multidimensional (power) distance analysis aka extended curvilinear (power) component analysis and extended curvilinear (power) distance analysis (Rusch, 2024, <doi:10.57938/355bf835-ddb7-42f4-8b85-129799fc240e>). Some functions are suitably flexible to allow any other sensible combination of explicit power transformations for weights, distances and input proximities with implicit ratio, interval, splines or nonmetric optimal scaling of the input proximities. Most functions use a Majorization-Minimization algorithm. Currently the methods are only available for one-mode two-way data (symmetric dissimilarity matrices).

This package performs variable selection based on subsampling, ranking forward selection. Details of the method are published in Lihui Liu, Hong Gu, Johan Van Limbergen, Toby Kenney (2020) SuRF: A new method for sparse variable selection, with application in microbiome data analysis Statistics in Medicine 40 897-919 <doi:10.1002/sim.8809>. Xo is the matrix of predictor variables. y is the response variable. Currently only binary responses using logistic regression are supported. X is a matrix of additional predictors which should be scaled to have sum 1 prior to analysis. fold is the number of folds for cross-validation. Alpha is the parameter for the elastic net method used in the subsampling procedure: the default value of 1 corresponds to LASSO. prop is the proportion of variables to remove in the each subsample. weights indicates whether observations should be weighted by class size. When the class sizes are unbalanced, weighting observations can improve results. B is the number of subsamples to use for ranking the variables. C is the number of permutations to use for estimating the critical value of the null distribution. If the doParallel package is installed, the function can be run in parallel by setting ncores to the number of threads to use. If the default value of 1 is used, or if the doParallel package is not installed, the function does not run in parallel. display.progress indicates whether the function should display messages indicating its progress. family is a family variable for the glm() fitting. Note that the glmnet package does not permit the use of nonstandard link functions, so will always use the default link function. However, the glm() fitting will use the specified link. The default is binomial with logistic regression, because this is a common use case. pval is the p-value for inclusion of a variable in the model. Under the null case, the number of false positives will be geometrically distributed with this as probability of success, so if this parameter is set to p, the expected number of false positives should be p/(1-p).

This package provides a comprehensive set of functions providing frequentist methods for network meta-analysis (Balduzzi et al., 2023) <doi:10.18637/jss.v106.i02> and supporting Schwarzer et al. (2015) <doi:10.1007/978-3-319-21416-0>, Chapter 8 "Network Meta-Analysis": - frequentist network meta-analysis following Rücker (2012) <doi:10.1002/jrsm.1058>; - additive network meta-analysis for combinations of treatments (Rücker et al., 2020) <doi:10.1002/bimj.201800167>; - network meta-analysis of binary data using the Mantel-Haenszel or non-central hypergeometric distribution method (Efthimiou et al., 2019) <doi:10.1002/sim.8158>, or penalised logistic regression (Evrenoglou et al., 2022) <doi:10.1002/sim.9562>; - rankograms and ranking of treatments by the Surface under the cumulative ranking curve (SUCRA) (Salanti et al., 2013) <doi:10.1016/j.jclinepi.2010.03.016>; - ranking of treatments using P-scores (frequentist analogue of SUCRAs without resampling) according to Rücker & Schwarzer (2015) <doi:10.1186/s12874-015-0060-8>; - split direct and indirect evidence to check consistency (Dias et al., 2010) <doi:10.1002/sim.3767>, (Efthimiou et al., 2019) <doi:10.1002/sim.8158>; - scatter plot to visualize local inconsistency (Wilson et al., 2026) <doi:10.1017/rsm.2026.10082>; - league table with network meta-analysis results; - comparison-adjusted funnel plot (Chaimani & Salanti, 2012) <doi:10.1002/jrsm.57>; - net heat plot and design-based decomposition of Cochran's Q according to Krahn et al. (2013) <doi:10.1186/1471-2288-13-35>; - measures characterizing the flow of evidence between two treatments by König et al. (2013) <doi:10.1002/sim.6001>; - automated drawing of network graphs described in Rücker & Schwarzer (2016) <doi:10.1002/jrsm.1143>; - partial order of treatment rankings ('poset') and Hasse diagram for poset (Carlsen & Bruggemann, 2014) <doi:10.1002/cem.2569>; (Rücker & Schwarzer, 2017) <doi:10.1002/jrsm.1270>; - contribution matrix as described in Papakonstantinou et al. (2018) <doi:10.12688/f1000research.14770.3> and Davies et al. (2022) <doi:10.1002/sim.9346>; - path-based approach for detecting and assessing inconsistency; - network meta-regression with a single continuous or binary covariate (Kwarteng et al., 2026) <doi:10.21203/rs.3.rs-8235913/v1>; - subgroup network meta-analysis.

This R package introduces Weighted Mean SHapley Additive exPlanations (WMSHAP), an innovative method for calculating SHAP values for a grid of fine-tuned base-learner machine learning models as well as stacked ensembles, a method not previously available due to the common reliance on single best-performing models. By integrating the weighted mean SHAP values from individual base-learners comprising the ensemble or individual base-learners in a tuning grid search, the package weights SHAP contributions according to each model's performance, assessed by multiple either R squared (for both regression and classification models). alternatively, this software also offers weighting SHAP values based on the area under the precision-recall curve (AUCPR), the area under the curve (AUC), and F2 measures for binary classifiers. It further extends this framework to implement weighted confidence intervals for weighted mean SHAP values, offering a more comprehensive and robust feature importance evaluation over a grid of machine learning models, instead of solely computing SHAP values for the best model. This methodology is particularly beneficial for addressing the severe class imbalance (class rarity) problem by providing a transparent, generalized measure of feature importance that mitigates the risk of reporting SHAP values for an overfitted or biased model and maintains robustness under severe class imbalance, where there is no universal criteria of identifying the absolute best model. Furthermore, the package implements hypothesis testing to ascertain the statistical significance of SHAP values for individual features, as well as comparative significance testing of SHAP contributions between features. Additionally, it tackles a critical gap in feature selection literature by presenting criteria for the automatic feature selection of the most important features across a grid of models or stacked ensembles, eliminating the need for arbitrary determination of the number of top features to be extracted. This utility is invaluable for researchers analyzing feature significance, particularly within severely imbalanced outcomes where conventional methods fall short. Moreover, it is also expected to report democratic feature importance across a grid of models, resulting in a more comprehensive and generalizable feature selection. The package further implements a novel method for visualizing SHAP values both at subject level and feature level as well as a plot for feature selection based on the weighted mean SHAP ratios.

This package implements a suite of semiparametric and nonparametric kernel-smoothed estimation and testing procedures for continuous mark-specific stratified hazard ratio (treatment/placebo) models in a randomized treatment efficacy trial with a time-to-event endpoint. Semiparametric methods, allowing multivariate marks, are described in Juraska M and Gilbert PB (2013), Mark-specific hazard ratio model with multivariate continuous marks: an application to vaccine efficacy. Biometrics 69(2):328-337 <doi:10.1111/biom.12016>, and in Juraska M and Gilbert PB (2016), Mark-specific hazard ratio model with missing multivariate marks. Lifetime Data Analysis 22(4):606-25 <doi:10.1007/s10985-015-9353-9>. Nonparametric kernel-smoothed methods, allowing univariate marks only, are described in Sun Y and Gilbert PB (2012), Estimation of stratified markâ specific proportional hazards models with missing marks. Scandinavian Journal of Statistics

An engine to facilitate the orchestration and execution of metadata-driven data management workflows, in compliance with FAIR (Findable, Accessible, Interoperable and Reusable) data management principles. By means of a pivot metadata model, relying on the DublinCore standard (<https://dublincore.org/>), a unique source of metadata can be used to operate multiple and inter-connected data management actions. Users can also customise their own workflows by creating specific actions but the library comes with a set of native actions targeting common geographic information and data management, in particular actions oriented to the publication on the web of metadata and data resources to provide standard discovery and access services. At first, default actions of the library were meant to focus on providing turn-key actions for geospatial (meta)data: 1) by creating manage geospatial (meta)data complying with ISO/TC211 (<https://committee.iso.org/home/tc211>) and OGC (<https://www.ogc.org/standards/>) geographic information standards (eg 19115/19119/19110/19139) and related best practices (eg. INSPIRE'); and 2) by facilitating extraction, reading and publishing of standard geospatial (meta)data within widely used software that compound a Spatial Data Infrastructure ('SDI'), including spatial databases (eg. PostGIS'), metadata catalogues (eg. GeoNetwork', CSW servers), data servers (eg. GeoServer'). The library was then extended to actions for other domains: 1) biodiversity (meta)data standard management including handling of EML metadata, and their management with DataOne servers, 2) in situ sensors, remote sensing and model outputs (meta)data standard management by handling part of CF conventions, NetCDF data format and OPeNDAP access protocol, and their management with Thredds servers, 3) generic / domain agnostic (meta)data standard managers ('DublinCore', DataCite'), to facilitate the publication of data within (meta)data repositories such as Zenodo (<https://zenodo.org>) or DataVerse (<https://dataverse.org/>). The execution of several actions will then allow to cross-reference (meta)data resources in each action performed, offering a way to bind resources between each other (eg. reference Zenodo DOI in GeoNetwork'/'GeoServer metadata, or vice versa reference GeoNetwork'/'GeoServer links in Zenodo or EML metadata). The use of standardized configuration files ('JSON or YAML formats) allow fully reproducible workflows to facilitate the work of data and information managers.

Set of tools for statistical analysis, visualization, and reporting of agroindustrial and agricultural experiments. The package provides functions to perform one-way and two-way ANOVA with post-hoc tests (Tukey HSD and Duncan MRT), Welch ANOVA for heteroscedastic data, and the Games-Howell post-hoc test as a robust alternative when variance homogeneity fails. Normality of residuals is assessed with the Shapiro-Wilk test and homoscedasticity with the Fligner-Killeen test; the appropriate statistical path is selected automatically based on these diagnostics. Coefficients of variation and statistical power (via one-way ANOVA power analysis) are reported alongside the post-hoc letter display. High-level wrappers allow automated multi-variable analysis with optional clustering by one or two experimental factors, with support for custom level ordering and relabeling. Results are returned as ggplot2 boxplots with mean and letter annotations, wide-format summary tables ready for publication or LaTeX rendering, and structured decision summaries for rapid agronomic interpretation. Direct export to Excel spreadsheets and high-resolution image tables is also supported. Functions follow methods widely used in agronomy, field trials, and plant breeding. Key references: Tukey (1949) <doi:10.2307/3001913>; Duncan (1955) <doi:10.2307/3001478>; Welch (1951) <doi:10.2307/2332579>; Games and Howell (1976) <doi:10.2307/2529858>; Shapiro and Wilk (1965) <doi:10.2307/2333709>; Fligner and Killeen (1976) <doi:10.2307/2529096>; Cohen (1988, ISBN:9781138892899); Wickham (2016, ISBN:9783319242750) for ggplot2'; see also agricolae <https://CRAN.R-project.org/package=agricolae> and rstatix <https://CRAN.R-project.org/package=rstatix>. Version en espanol: Conjunto de herramientas para el analisis estadistico, visualizacion y generacion de reportes en ensayos agroindustriales y agricolas. Incluye ANOVA univariado y bifactorial con pruebas post-hoc (Tukey HSD y Duncan MRT), ANOVA de Welch para datos heterocedasticos y la prueba post-hoc de Games-Howell como alternativa robusta cuando falla la homogeneidad de varianzas. La normalidad de residuos se evalua con la prueba de Shapiro-Wilk y la homogeneidad de varianzas con la prueba de Fligner-Killeen; la ruta estadistica apropiada se selecciona automaticamente segun estos diagnosticos. Se reportan coeficientes de variacion y potencia estadistica junto con las letras de separacion de medias. Los envoltorios de alto nivel permiten analisis multivariable automatizado con agrupamiento opcional por uno o dos factores experimentales, con soporte para orden y etiquetado personalizado de niveles. Los resultados se devuelven como boxplots con anotaciones de medias y letras, tablas resumen en formato ancho listas para publicacion o renderizado en LaTeX, y resumenes de decision para interpretacion agronomica rapida. Tambien se soporta exportacion directa a Excel e imagenes de alta resolucion para informes tecnicos.

This package provides a fully native R deep learning framework for constructing, training, evaluating, and inspecting Deep Dynamic Ensemble Self Organizing Neural Networks at research scale. The core engine is an object oriented R6 class-based implementation with explicit control over layer layout, dimensional flow, forward propagation, back propagation, and transparent optimizer state updates. The framework does not rely on external deep learning back ends, enabling direct inspection of model state, reproducible numerical behavior, and fine grained architectural control without requiring compiled dependencies or graphics processing unit specific run times. Users can define dimension agnostic single layer or deep multi-layer networks without hard coded architecture limits, with per layer configuration vectors for activation functions, derivatives, dropout behavior, and initialization strategies automatically aligned to network depth through controlled replication or truncation. Reproducible workflows can be executed through high level helpers for fit, run, and predict across binary classification, multi-class classification, and regression modes. Training pipelines support optional self organization, adaptive learning rate behavior, and structured ensemble orchestration in which candidate models are evaluated under user specified performance metrics and selectively promoted or pruned to refine a primary ensemble, enabling controlled ensemble evolution over successive runs. Ensemble evaluation includes fused prediction strategies in which member outputs may be combined through weighted averaging, arithmetic averaging, or voting mechanisms to generate consolidated metrics for research level comparison and reproducible per-seed assessment. The framework supports multiple optimization approaches, including stochastic gradient descent, adaptive moment estimation, and look ahead methods, alongside configurable regularization controls such as L1, L2, and mixed penalties with separate weight and bias update logic. Evaluation features provide threshold tuning, relevance scoring, receiver operating characteristic and precision recall curve generation, area under curve computation, regression error diagnostics, and report ready metric outputs. The package also includes artifact path management, debug state utilities, structured run level metadata persistence capturing seeds, configuration states, thresholds, metrics, ensemble transitions, fused evaluation artifacts, and model identifiers, as well as reproducible scripts and vignettes documenting end to end experiments. Kingma and Ba (2015) <doi:10.48550/arXiv.1412.6980> "Adam: A Method for Stochastic Optimization". Hinton et al. (2012) <https://www.cs.toronto.edu/~tijmen/csc321/slides/lecture_slides_lec6.pdf> "Neural Networks for Machine Learning (RMSprop lecture notes)". Duchi et al. (2011) <https://jmlr.org/papers/v12/duchi11a.html> "Adaptive Subgradient Methods for Online Learning and Stochastic Optimization". Zeiler (2012) <doi:10.48550/arXiv.1212.5701> "ADADELTA: An Adaptive Learning Rate Method". Zhang et al. (2019) <doi:10.48550/arXiv.1907.08610> "Lookahead Optimizer: k steps forward, 1 step back". You et al. (2019) <doi:10.48550/arXiv.1904.00962> "Large Batch Optimization for Deep Learning: Training BERT in 76 minutes (LAMB)". McMahan et al. (2013) <https://research.google.com/pubs/archive/41159.pdf> "Ad Click Prediction: a View from the Trenches (FTRL-Proximal)". Klambauer et al. (2017) <https://proceedings.neurips.cc/paper/6698-self-normalizing-neural-networks.pdf> "Self-Normalizing Neural Networks (SELU)". Maas et al. (2013) <https://ai.stanford.edu/~amaas/papers/relu_hybrid_icml2013_final.pdf> "Rectifier Nonlinearities Improve Neural Network Acoustic Models (Leaky ReLU / rectifiers)".

tools for building book.

External jars required for package RKEA.

Various databases of microRNA Targets.

Bootstrap, permutation tests, and jackknife, featuring easy-to-use syntax.

Three-step regression and inference for cellwise and casewise contamination.

Administrative regions and other spatial objects of the Czech Republic.