Skip to content

Latest commit

 

History

History
161 lines (105 loc) · 5.51 KB

NEWS.md

File metadata and controls

161 lines (105 loc) · 5.51 KB
Raw

domir 1.2.0

visible

  • Basic parallel estimation of all subsets using parallel::parSapply
  • Progress bar added for non-parallel sub-model estimation
  • Complete dominance reports fraction of total models higher/lower
    • Update to summary method
  • offset()s now allowed with formula method
  • formula input with .adj defunct
  • Better handing of errors returned by '.fct'. Reports the specific error that resulted.
  • New pkgdown-based GitHub page with extra visual design features.

internal

  • formula method sync with formula_list
  • Intercept removal and offsets not allowed in .all and .set

bug fix

  • offsets() omitted from models with no predictors. Was an issue with .adj's value

domir 1.1.1

bug fixes

  • domir() produced error when .cdl = FALSE was combined with .all
  • lme4 added to suggested list due to reference in documentation and CRAN warning

domir 1.1.0

visible

  • formula_list method implemented
    • merger of planned list and Formula::Formula methods
    • fmllst2Fml can be used to translate formula_lists to Formula::Formulas
    • offset()s allowed in formula_list; to update in formula method
  • formula method .adj argument is logical
    • formulas of the form ~ 1 still accepted but depreciated
    • other formula inputs defunct and user pushed to add to .fct

internal

  • Began linting process for code readability

domir 1.0.1

bug fixes

  • formula method in domir() and domin() now disallow offset() terms
    • Were silently removed and not replaced during parsing.
  • Argument checks for formula method produced inconsistent results

domir 1.0.0

breaking/major changes

  • New domir function
    • Generic function with three S3 methods: formula, Formula, and list
      • Formula and list methods under development
    • .obj submits object on which S3 method dispatch is based
    • .fct submits function that accepts result of same type as .obj and returns a scalar
    • .set, .all and .adj affect .obj entries
      • .wst argument under development ('within-set' or Owen decomposition)
  • domin superseded
    • No longer under active development; bug fixes only

internal

  • dominance_scalar internal function responsible for computation. domir is front-end to computational engine.
    • Other computational engines planned. Internal function may be renamed.

bug fixes

  • summary.domin not recognizing reverse = TRUE.

domir 0.3.2

bug fixes

  • removing reference to dominanceanalysis package not on CRAN currently given request from CRAN maintainers and CRAN note.

domir 0.3.1

visible

  • fixing typos and other editorial errors in domir_basics vignette.
  • Extending conditional dominance discussion with graphic.

bug fixes

  • fixing error in email address in domir-package leading to CRAN check warnings.

domir 0.3.0

visible

  • conditional logical argument to make computing conditional dominance statistics optional
  • initial working version of summary.domin
    • method for computing the strongest dominance designations between independent variables
    • update to print.domin to display strongest dominance designations
  • reorganization and update to Conceptual Introduction to Dominance Analysis vignette

internal

  • looping used to construct many dominance statistics, replaced with matrix-based methods; intended to simplify and expedite computations

bug fixes

  • update to email in domin-package as requested by CRAN maintainers.

domir 0.2.0

visible

  • consmodel argument to adjust for baseline fit statistic values
  • reverse argument to change rank and complete dominance interpretation for fit metrics that decrease with better fit
  • Conceptual Introduction to Dominance Analysis vignette added

internal

  • implemented unit testing framework - covers 82% of domin
  • internal use of reformulate to construct formulas as opposed to paste-ing +-es

bug fixes

  • Complete_Dominance computation producing inconsistent complete dominance designations - within-order complete dominance checks not capturing all possible model comparisons

domir 0.1.0

visible

  • update to format of Complete_Dominance from integers to logicals
  • output and warnings no longer suppressed by default - user must silence noisy functions
  • minimum of two terms/sets to run domin (replicates behavior of Stata version)
  • additional checks (response in formula/orders in formula)
  • names of entries in domin object changed to syntactic R names when not. Affects entries in:
    • Complete_Dominance matrix
    • Conditional_Dominance matrix ## internal
  • extensive re-write of computation methods
  • pre-allocates container objects to improve performance
  • increased functional-ization of subroutines and overall code readability
  • many internal functions re-named

bug fixes

  • returned value .$Subset_Details$Full_Model now includes variables in all
  • Complete_Dominance matrix computation
    • Too many models considered for complete dominance - inconsistent with standard dominance analysis methodology
    • Fixed error in suppressing complete dominance (i.e., complete = FALSE) resulted in error

domir 0.0.1

bug fixes

  • Ensemble_Coordinator that submitted factor numbers instead of text in R versions < 4.0

domir 0.0.0

  • initial working version of domin
  • initial working version of print.domin