Native R tools for optimal binning workflows in predictive modeling. The package provides APIs for binary, multi-class and continuous targets, with multi-variable binning and scorecard workflows. Methods are informed by Navas-Palencia (2020) <doi:10.48550/arXiv.2001.08025> and Navas-Palencia (2021) <doi:10.48550/arXiv.2104.08619>.

This tool was designed to assess the sensitivity of research findings to omitted variables when estimating causal effects using propensity score (PS) weighting. This tool produces graphics and summary results that will enable a researcher to quantify the impact an omitted variable would have on their results. Burgette et al. (2021) describe the methodology behind the primary function in this package, ov_sim. The method is demonstrated in Griffin et al. (2020) <doi:10.1016/j.jsat.2020.108075>.

Introduces optional types with some() and none, as well as match_with() from functional languages.

Optimal group-sequential designs minimise some function of the expected and maximum sample size whilst controlling the type I error rate and power at a specified level. OptGS provides functions to quickly search for near-optimal group-sequential designs for normally distributed outcomes. The methods used are described in Wason, JMS (2015) <doi:10.18637/jss.v066.i02>.

Solver for linear, quadratic, and rational programs with linear, quadratic, and rational constraints. A unified interface to different R packages is provided. Optimization problems are transformed into equivalent formulations and solved by the respective package. For example, quadratic programming problems with linear, quadratic and rational constraints can be solved by augmented Lagrangian minimization using package alabama', or by sequential quadratic programming using solver slsqp'. Alternatively, they can be reformulated as optimization problems with second order cone constraints and solved with package cccp'.

An implementation for computing Optimal B-Robust Estimators of two-parameter distribution. The procedure is composed of some equations that are evaluated alternatively until the solution is reached. Some tools for analyzing the estimates are included. The most relevant is covariance matrix computation using a closed formula.

R bindings to odiff', a blazing-fast pixel-by-pixel image comparison tool <https://github.com/dmtrKovalenko/odiff>. Supports PNG, JPEG, WEBP, and TIFF with configurable thresholds, antialiasing detection, and region ignoring. Requires system installation of odiff'. Ideal for visual regression testing in automated workflows.

This package provides functions to download and tidy statistical data published by the Office for National Statistics <https://www.ons.gov.uk>. Covers GDP, inflation (CPI, CPIH, RPI), unemployment, employment, wages, trade, retail sales, house prices, productivity, population, and public sector finances. Most series are fetched from the ONS website using its CSV time series endpoint. House price data is sourced from HM Land Registry <https://www.gov.uk/government/organisations/land-registry>. Data is cached locally between sessions.

Takes images, imported via imager', and converts them into a data frame that can be plotted to look like the imported image. This can be used for creating data that looks like a specific image. Additionally, images with color and alpha channels can be converted to grayscale in preparation for converting to the data frame format.

This package provides functions for creating ensembles of optimal trees for regression, classification (Khan, Z., Gul, A., Perperoglou, A., Miftahuddin, M., Mahmoud, O., Adler, W., & Lausen, B. (2019). (2019) <doi:10.1007/s11634-019-00364-9>) and class membership probability estimation (Khan, Z, Gul, A, Mahmoud, O, Miftahuddin, M, Perperoglou, A, Adler, W & Lausen, B (2016) <doi:10.1007/978-3-319-25226-1_34>) are given. A few trees are selected from an initial set of trees grown by random forest for the ensemble on the basis of their individual and collective performance. Three different methods of tree selection for the case of classification are given. The prediction functions return estimates of the test responses and their class membership probabilities. Unexplained variations, error rates, confusion matrix, Brier scores, etc. are also returned for the test data.

Defines thresholds for breaking data into a number of discrete levels, minimizing the (mean) squared error within all bins.

In high-dimensional streaming data analysis, extracting core periodic features under real-time constraints remains challenging. Traditional dimension reduction methods fail to adapt to incremental data and yield low accuracy due to irrelevant variables. This package provides the Online Sliced Inverse Regression framework for cosine regression with high-dimensional irrelevant variables. It integrates subspace extraction of sliced inverse regression and incremental learning of online algorithms to efficiently handle periodic streaming data. Cai, Z., Li, R., & Zhu, L. (2020) <doi:10.48550/arXiv.2002.02795>.

An interface to easily run local language models with Ollama <https://ollama.com> server and API endpoints (see <https://github.com/ollama/ollama/blob/main/docs/api.md> for details). It lets you run open-source large language models locally on your machine.

This package implements ordered beta regression models, which are for modeling continuous variables with upper and lower bounds, such as survey sliders, dose-response relationships and indexes. For more information, see Kubinec (2023) <doi:10.31235/osf.io/2sx6y>. The package is a front-end to the R package brms', which facilitates a range of regression specifications, including hierarchical, dynamic and multivariate modeling.

This package provides a generalised data structure for fast and efficient loading and data munching of sparse omics data. The OmicFlow requires an up-front validated metadata template from the user, which serves as a guide to connect all the pieces together by aligning them into a single object that is defined as an omics class. Once this unified structure is established, users can perform manual subsetting, visualisation, and statistical analysis, or leverage the automated autoFlow method to generate a comprehensive report.

Detect the number and locations of change points. The locations can be either exact or in terms of ranges, depending on the available computational resource. The method is based on Jie Ding, Yu Xiang, Lu Shen, Vahid Tarokh (2017) <doi:10.1109/TSP.2017.2711558>.

This package provides a derivative-based optimization framework that allows users to combine eight convergence criteria. Unlike standard optimization functions, this package includes a built-in mechanism to verify the positive definiteness of the Hessian matrix at the point of convergence. This additional check helps prevent the solver from falsely identifying non-optimal solutions, such as saddle points, as valid minima.

In bulk epigenome/transcriptome experiments, molecular expression is measured in a tissue, which is a mixture of multiple types of cells. This package tests association of a disease/phenotype with a molecular marker for each cell type. The proportion of cell types in each sample needs to be given as input. The package is applicable to epigenome-wide association study (EWAS) and differential gene expression analysis. Takeuchi and Kato (submitted) "omicwas: cell-type-specific epigenome-wide and transcriptome association study".

Obtain and evaluate various optimal designs for the 3, 4, and 5-parameter logistic models. The optimal designs are obtained based on the numerical algorithm in Hyun, Wong, Yang (2018) <doi:10.18637/jss.v083.i05>.

This package provides a database resource that is accessible through the Open Database Connectivity ('ODBC') API. This package uses the Resource model, with URL "resolver" and "client", to dynamically discover and make accessible tables stored in a MS SQL Server database. For more details see Marcon (2021) <doi:10.1371/journal.pcbi.1008880>.

Fits community site occupancy models to environmental DNA metabarcoding data collected using spatially-replicated survey design. Model fitting results can be used to evaluate and compare the effectiveness of species detection to find an efficient survey design. Reference: Fukaya et al. (2022) <doi:10.1111/2041-210X.13732>, Fukaya and Hasebe (2025) <doi:10.1002/1438-390X.12219>.

Identifies the optimal transformation of a surrogate marker and estimates the proportion of treatment explained (PTE) by the optimally-transformed surrogate at an earlier time point when the primary outcome of interest is a censored time-to-event outcome; details are described in Wang et al (2021) <doi:10.1002/sim.9185>.

This package provides analyse, interpret and understand noise pollution data. Data are typically regular time series measured with sound meter. The package is partially described in Fogola, Grasso, Masera and Scordino (2023, <DOI:10.61782/fa.2023.0063>).

Estimates out-of-sample R² through bootstrap or cross-validation as a measure of predictive performance. In addition, a standard error for this point estimate is provided, and confidence intervals are constructed.