Automatic Fixed Rank Kriging

Automatic fixed rank kriging for (irregularly located) spatial data using a class of basis functions with multi-resolution features and ordered in terms of their resolutions. The model parameters are estimated by maximum likelihood (ML) and the number of basis functions is determined by Akaike's information criterion (AIC). For spatial data with either one realization or independent replicates, the ML estimates and AIC are efficiently computed using their closed-form expressions when no missing value occurs. Details regarding the basis function construction, parameter estimation, and AIC calculation can be found in Tzeng and Huang (2018) . For data with missing values, the ML estimates are obtained using the expectation- maximization algorithm. Apart from the number of basis functions, there are no other tuning parameters, making the method fully automatic. Users can also include a stationary structure in the spatial covariance, which utilizes 'LatticeKrig' package.

Robust Clustering Procedures

A clustering algorithm similar to K-Means is implemented, it has two main advantages, namely (a) The estimator is resistant to outliers, that means that results of estimator are still correct when there are atypical values in the sample and (b) The estimator is efficient, roughly speaking, if there are no outliers in the sample, results will be similar to those obtained by a classic algorithm (K-Means). Clustering procedure is carried out by minimizing the overall robust scale so-called tau scale. (see Gonzalez, Yohai and Zamar (2019) ).

Versatile Functions for Working with Pedigrees

Tools to sort, edit and prune pedigrees and to extract the inbreeding coefficients and the relationship matrix (includes code for pedigrees from self-pollinated species). The use of pedigree data is central to genetics research within the animal and plant breeding communities to predict breeding values. The relationship matrix between the individuals can be derived from pedigree structure ('Vazquez et al., 2010') .

Cytokine Profiling Analysis Tool

Provides comprehensive cytokine profiling analysis through quality control using biologically meaningful cutoffs on raw cytokine measurements and by testing for distributional symmetry to recommend appropriate transformations. Offers exploratory data analysis with summary statistics, enhanced boxplots, and barplots, along with univariate and multivariate analytical capabilities for in-depth cytokine profiling such as Principal Component Analysis based on Andrzej Maćkiewicz and Waldemar Ratajczak (1993) , Sparse Partial Least Squares Discriminant Analysis based on Lê Cao K-A, Boitard S, and Besse P (2011) , Random Forest based on Breiman, L. (2001) , and Extreme Gradient Boosting based on Tianqi Chen and Carlos Guestrin (2016) .

Portfolio Analysis, Including Numerical Methods for Optimization of Portfolios

Portfolio optimization and analysis routines and graphics.

Impute Missing Glucose Values in CGM Data

Imputes missing glucose values in repeated-measures continuous glucose monitoring (CGM) data. Workflows create time-series features from raw timestamps, support model selection, and return the user's original columns plus an imputed glucose column. Methods include multiple imputation by chained equations (MICE; Azur et al. (2011) ), Random Forest regression (Breiman (2001) ), k-nearest-neighbor regression (Zhang (2016) ), XGBoost (Chen and Guestrin (2016) ), LightGBM (Ke et al. (2017) < https://papers.nips.cc/paper/6907-lightgbm-a-highly-efficient-gradient-boosting-decision>), and ARIMA forecasting with the forecast framework (Hyndman and Khandakar (2008) ). A Python-compatible backend uses 'reticulate' to call 'pandas', 'scikit-learn', 'statsmodels', Python 'xgboost', and optional Python 'lightgbm'.

Impute Missing Glucose Values in CGM Data

Imputes missing glucose values in repeated-measures continuous glucose monitoring (CGM) data. Workflows create time-series features from raw timestamps, support model selection, and return the user's original columns plus an imputed glucose column. Methods include multiple imputation by chained equations using 'mice' (Azur et al. (2011) ), Random Forest regression using 'ranger' (Breiman (2001) ), k-nearest-neighbor regression using 'FNN' (Zhang (2016) ), 'XGBoost' using 'xgboost' (Chen and Guestrin (2016) ), 'LightGBM' using 'lightgbm' (Ke et al. (2017) < https://papers.nips.cc/paper/6907-lightgbm-a-highly-efficient-gradient-boosting-decision>), and ARIMA forecasting using 'forecast' (Hyndman and Khandakar (2008) ). A 'Python'-compatible backend uses 'reticulate' to call 'pandas', 'scikit-learn', 'statsmodels', 'xgboost', and optional 'lightgbm'.

Exploratory Reduced Reparameterized Unified Model Estimation

Perform a Bayesian estimation of the exploratory reduced reparameterized unified model (ErRUM) described by Culpepper and Chen (2018) .

Bayesian Estimation of an Exploratory Deterministic Input, Noisy and Gate Model

Perform a Bayesian estimation of the exploratory deterministic input, noisy and gate (EDINA) cognitive diagnostic model described by Chen et al. (2018) .

Estimating Infection Rates from Serological Data

Translates antibody levels measured in cross-sectional population samples into estimates of the frequency with which seroconversions (infections) occur in the sampled populations. Replaces the previous `seroincidence` package.