Initial public development version. This is a pre-release: the API may change before 1.0.0. The largest changes are expected in the Category C methods (signal-difference regions, peak evidence, calibration transfer risk, kinetic discrimination, experiment fragility, drift uncertainty, detection reliability, and uncertainty-origin attribution), where the interface is still settling. Version pinning supports reproducible analyses.
32 user-facing functions, grouped into three categories.
sci_describe - descriptive statistics with explicit biological vs. technical replicate handlingsci_compare - two-group comparisons (Welch, Student, paired, permutation, Wilcoxon) with effect sizessci_anova - ANOVA and Welch alternatives with effect sizessci_regression - linear, weighted, and robust regression with diagnosticssci_correlation - correlation analysis with confidence intervals (Pearson, Spearman, Kendall)sci_normality - normality assessment with multiple testssci_outliers - outlier diagnostics (IQR, MAD, Grubbs, Dixon)sci_power - power and sample-size calculation for common designssci_effect_size - standardized effect sizes (Cohen's d, Hedges' g) with CIssci_bootstrap - bootstrap resampling with percentile, BCa, and normal CIsadaptive_baseline - asymmetric least squares / lowess baseline estimation for spectraadaptive_smoothing - edge-preserving smoothing for noisy signalsbatch_effect_quantify - quantify batch-to-batch variance via mixed-effects decompositioncalibration_compare - compare two calibration curves via ANCOVAcurve_similarity - similarity between two curves sampled on possibly different gridsdrift_correct_stat - drift correction with uncertainty quantificationkinetic_model_compare - compare candidate kinetic models via likelihood-ratio testsnoise_characterize - summarise residual variation and autocorrelationparameter_uncertainty - propagate parameter uncertainty to derived quantitiesreplicate_decompose - variance-component decomposition for nested and crossed replicatesrobust_lod - limits of detection and quantification (Currie-style)spectral_change_test - test for differences across complete spectratimepoint_optimize - optimise sampling timepoints for a planned experimentweighted_peak_fit - weighted multi-peak fitting (Gaussian/Lorentzian)signal_difference_regions - statistical identification of regions where two groups of curves differ (simultaneous max-t band with Westfall-Young permutation)peak_evidence - frequentist evidence assessment for signal peaks (likelihood ratio + permutation FPP)calibration_transfer_risk - statistical framework for calibration transfer risk (TPE, OOD flag, parametric bootstrap)kinetic_discrimination - discriminability of competing kinetic models via Monte Carlo simulationexperiment_fragility - quantitative experimental fragility assessment (leave-one-replicate-out, outlier, baseline, model, noise)drift_uncertainty - statistical attribution of variance to instrument drift (linear/polynomial/exponential drift models)detection_reliability - reliability analysis for analytical detection (PD/FNR/FPR curves with bootstrap CIs)uncertainty_origin - stage-wise uncertainty attribution via Shapley value for correlated sourcescrossstatr_result parent class with print, summary, tidy, as.data.frame methods.assert_*) and numeric helpers (.safe_sd, .ci_t, .cohens_d, .mcse_p, etc.)testthat (3rd edition) test suite covering input validation, numerical correctness, and edge casesinst/CITATION, inst/extdata/ for example data