A Procedure to Clean, Decompose and Aggregate Timeseries

Clean, decompose and aggregate univariate time series following the procedure "Cyclic/trend decomposition using bin interpolation" and the Logbox method for flagging outliers, both detailed in Ritter, F.: Technical note: A procedure to clean, decompose, and aggregate time series, Hydrol. Earth Syst. Sci., 27, 349–361, , 2023.

Functions to Assess the Business Impact of Churn Prediction Models

Calculate and visualise the financial impact of using a classification model, such as a churn model, to target customers. Provides cost, revenue, profit and return-on-investment curves as a function of the share of customers targeted, cumulative gains and lift, marginal profit per bin, and confusion-matrix based payoff across probability thresholds. Also includes 'ggplot2' 'autoplot()' methods and an interactive 'shiny' application for exploring the results.

Collapses Levels, Computes Information Value and WoE

Contains functions to help in selecting and exploring features ( or variables ) in binary classification problems. Provides functions to compute and display information value and weight of evidence (WoE) of the variables , and to convert numeric variables to categorical variables by binning. Functions are also provided to determine which levels ( or categories ) of a categorical variable can be collapsed (or combined ) based on their response rates. The functions provided only work for binary classification problems.

Rank and Social Hierarchy for Gregarious Animals

Tools developed to facilitate the establishment of the rank and social hierarchy for gregarious animals by the Si method developed by Kondo & Hurnik (1990). It is also possible to determine the number of agonistic interactions between two individuals, sociometric and dyadics matrix from dataset obtained through electronic bins. In addition, it is possible plotting the results using a bar plot, box plot, and sociogram.

Estimate Bunching

Implementation of the bunching estimator for kinks and notches. Allows for flexible estimation of counterfactual (e.g. controlling for round number bunching, accounting for other bunching masses within bunching window, fixing bunching point to be minimum, maximum or median value in its bin, etc.). It produces publication-ready plots in the style followed since Chetty et al. (2011) , with lots of functionality to set plot options.

Dynamic Programming Based Gaussian Mixture Modelling Tool for 1D and 2D Data

Gaussian mixture modeling of one- and two-dimensional data, provided in original or binned form, with an option to estimate the number of model components. The method uses Gaussian Mixture Models (GMM) with initial parameters determined by a dynamic programming algorithm, leading to stable and reproducible model fitting. For more details see Zyla, J., Szumala, K., Polanski, A., Polanska, J., & Marczyk, M. (2026) .

Scorecard Development and Internal Ratings-Based Risk Parameters

Builds points scorecards for binary targets (credit risk, fraud, propensity) on the optimal binning and weight of evidence engine of 'OptimalBinningWoE', and takes them to the risk parameters of the internal ratings-based (IRB) approach. Variables are selected through optimal binning, eight admission rules, hold-out revalidation with frozen bins and a consensus of 'glmnet', 'xgboost', 'lightgbm' and 'ranger' models weighted by out-of-sample performance; the audit funnel never drops a candidate from the report. The scorecard is fitted with an explicit, auditable scale alignment (a log-odds regression on the raw score composed with the points-to-double-the-odds map); cut-offs are swept with frozen cuts; reject inference is reported as a sensitivity band; the population and characteristic stability indices (PSI and CSI) are monitored with both the fixed and the sample-size-adjusted threshold; and production SQL is generated in fourteen dialects, with the agreement between R and SQL verified by test. The IRB layer builds the default flag; calibrates the scorecard to a long-run default rate with rating grades, margins of conservatism and floors to give the probability of default (PD); models workout loss given default (LGD) in two stages with downturn and in-default estimates; models credit conversion factors from facility snapshots to give the exposure at default (EAD); and computes expected loss, risk weights, regulatory capital and expected credit loss from parameter tables selected by framework preset. The heavy numeric kernels (rank correlation of wide weight of evidence tables, exact concordance counts for Somers' D, streamed expected credit loss paths) are compiled with 'RcppArmadillo'. The scorecard methodology follows Siddiqi (2017) and Thomas et al. (2017) .

Fast 3D Lacunarity for Voxel Data

Calculates 3D lacunarity from voxel data. It is designed for use with point clouds generated from Light Detection And Ranging (LiDAR) scans in order to measure the spatial heterogeneity of 3-dimensional structures such as forest stands. It provides fast 'C++' functions to efficiently bin point cloud data into voxels and calculate lacunarity using different variants of the gliding-box algorithm originated by Allain & Cloitre (1991) .

Tools for Poisson Data

Functions used for analyzing count data, mostly crime counts. Includes checking difference in two Poisson counts (e-test), checking the fit for a Poisson distribution, small sample tests for counts in bins, Weighted Displacement Difference test (Wheeler and Ratcliffe, 2018) , to evaluate crime changes over time in treated/control areas. Additionally includes functions for aggregating spatial data and spatial feature engineering.

Spline Based Window Boundaries for Genomic Analyses

Defines window or bin boundaries for the analysis of genomic data. Boundaries are based on the inflection points of a cubic smoothing spline fitted to the raw data. Along with defining boundaries, a technique to evaluate results obtained from unequally-sized windows is provided. Applications are particularly pertinent for, though not limited to, genome scans for selection based on variability between populations (e.g. using Wright's fixations index, Fst, which measures variability in subpopulations relative to the total population).