Changes in version 2.2.0 (2026-01-08)                  

  - new selection criteria for unsupervised eigenvector selection via
    stepwise regression:
      - selection based on the corrected Akaike information criterion
        ('AICc') now available in glmFilter()
      - lmFilter() now supports eigenvector selection based on AIC,
        AICc, and BIC
  - lasso-based eigenvector selection now supported in lmFilter()
  - lmFilter() now allows to compute conditional standard errors for
    regression coefficients using a partial regression framework.
  - minor adjustment to the summary method
  - add deviance residuals for negative binomial regression models
  - add warning message that deviance residuals will become the default
    in glmFilter() in future releases
  - update vignette & tests

                 Changes in version 2.1.0 (2025-02-21)                  

  - update vignette to include an example of the negative binomial model
  - imrprovement in the console output of function vp()
  - improvements and bug fixes in MI.vec() and MI.decomp():
      - correctly handle missing values in each variable separately if
        multiple variables are supplied
      - check for variable names only inside the function and not in the
        global environment
      - removal of constant terms supplied to the functions
  - improve the handling of missingness in MI.resid()
  - minor adjustments to helper functions
  - update tests

                 Changes in version 2.0.0 (2025-01-22)                  

  - allow for unsupervised eigenvector selection in negative binomial
    models
      - glmFilter() now supports 'nb' (for negative binomial) as model
        type
      - adjustments in summary method and helper functions to handle
        negative binomial models
      - update tests for negative binomial model
  - glmFilter() also provides McFadden's adjusted pseudo R-squared for
    the filtered vs. the unfiltered model
  - bug fixes
      - bug fix in help pages of MI.local() and MI.vec() functions
      - resolves an error in lmFilter() and glmFilter() occurring when
        covariates are supplied as data.frame
  - assign variable names to output (if provided)
  - improve the handling of missingness in MI.vec(), MI.decomp(), and
    MI.local()
  - update tests

                 Changes in version 1.1.5 (2022-08-22)                  

  - fix minor bug in help pages

                 Changes in version 1.1.4 (2022-03-09)                  

  - update citation information
  - fix: use isTRUE(all.equal()) instead of "==" on numeric vectors

                        Changes in version 1.1.3                        

  - update tests

                 Changes in version 1.1.2 (2022-01-18)                  

  - fix broken links
  - update citation information

                 Changes in version 1.1.1 (2021-10-05)                  

  - CRAN resubmission
  - improve readability of code
  - update author mail address
  - include citation

                     Changes in version 1.1.0.9000                      

  - improve readability of code
  - update author mail address
  - include citation

                 Changes in version 1.1.0 (2021-05-12)                  

  - CRAN resubmission
  - fix minor bug when checking 'tol' in lmFilter() function
  - new vignette name
  - add new functions
      - MI.local() function to calculate local Moran's I
      - vp() function for variation partitioning
  - update reference to MI.local() in documentation files

                     Changes in version 1.0.0.9000                      

  - fix minor bug when checking 'tol' in lmFilter() function
  - new vignette name
  - add new functions
      - MI.local() function to calculate local Moran's I
      - vp() function for variation partitioning
  - update reference to MI.local() in documentation files

                 Changes in version 1.0.0 (2020-12-07)                  

  - include MI.decomp() to decompose Moran's I
  - rename MI.resid()
  - prepare for CRAN submission

                        Changes in version 0.2.0                        

  - lmFilter() and glmFilter() now also support unsupervised eigenvector
    selection based on the significance of residual autocorrelation

                        Changes in version 0.1.0