NEWS.md

In sharp: Stability-enHanced Approaches using Resampling Procedures

sharp version 1.4.7

• Add reference to the publication in the Journal of Statistical Software

sharp version 1.4.6

• Update the vignette

sharp version 1.4.5

• Allow for alternative optimisation methods implemented in nloptr
• Update parallelisation, now using the future package
• Fix the formatting of continuous outcome in VariableSelection()
• Update the vignette

sharp version 1.4.4

• Update references with published articles

sharp version 1.4.3

• Add sparse K means from the R package sparcl
• Allow for missing values in proportions for more flexibility

sharp version 1.4.2

• Remove functions depending on regsem (removed from CRAN)
• Fix the use of packages in Suggests in the examples

sharp version 1.4.1

• Add package vignette
• Use Ridge regression calibrated by cross validation instead of unpenalised regression in Refit(), ExplanatoryPerformance() and Incremental()
• Add new S3 class structural_model
• Fix inclusion of unpenalised predictors in Incremental()
• Fix clustering of rows in Clustering()

sharp version 1.4.0

• Update the stability score used by default (n_cat=NULL), previous score can be used with n_cat=3
• Add new functions for structural equation modelling including StructuralModel(), PenalisedSEM(), PenalisedOpenMx(), PenalisedLinearSystem(), LavaanModel(), LavaanMatrix(), OpenMxModel(), OpenMxMatrix() and LinearSystemMatrix()
• Add new function CART() for classification and regression trees
• Add the option to run randomised or adaptive lasso in PenalisedRegression()
• Fix a bug when running multinomial lasso with predictors with null variance in the subsamples
• Fix a bug where additional parameters in ... were used in glm.control() within Refit()

sharp version 1.3.0

• Add new functions for consensus clustering including Clustering(), Clusters(), ConsensusMatrix(), ClusteringPerformance() and more
• Add new print(), plot() and summary() functions
• Update plotting functions
• Fix parallelisation using argument n_cores in main functions
• Remove duplicated messages in ExplanatoryPerformance()
• Allow for factor ydata in VariableSelection() and related functions

sharp version 1.2.1

• Update examples for use with fake 1.3.0
• Fix requirements on input data format in Refitting()
• Add resampling argument in Explanatory()
• Add optional beep at the end of the run in main functions
• Increase igraph vertex size in Graph() and plot()

sharp version 1.2.0

• Add the functions Ensemble() and EnsemblePredictions() to build and predict from an ensemble model for VariableSelection()
• Add S3 classes including coef() and predict() for VariableSelection()
• Rename Recalibrate() as Refit()
• Fix use of CPSS in GraphicalModel()
• Fix maximisation of the contrast
• Add simulation functions to the companion R package fake

sharp version 1.1.0

First release of stability selection methods and simulation models.

sharp documentation built on April 11, 2025, 5:44 p.m.