v0.6.0: CRAN Release

midr 0.6.0 Release Notes (2026-03-08)

This release focuses on expanding support for multiple models and multivariate responses, optimizing core algorithm performance, and streamlining class and argument names.

1. Major Features & Enhancements

• Multiple Models & Multivariate Responses ("midlist" and "midrib" classes): Implemented collection objects of class "midlist" and "midrib", along with manipulation functions (e.g., midlist()) and corresponding visualization tools (methods for ggmid() and plot()). This enables the efficient handling and comparison of multiple interpretation models.

• Multi-class & Survival Model Interpretation: interpret() now accepts a matrix for the response variable y, allowing for the simultaneous modeling of multiple responses. This facilitates the interpretation of both multi-class classification models and survival models.

• Parametric Link Functions: Added support for parametric link functions including Box-Cox, Yeo-Johnson, Robit, and Scobit.

• Variable Encoding Updates: Enhanced the variable encoding functionality in interpret() and encoder generators by introducing new arguments: split for numeric variables and lump for factor variables.

2. Breaking Changes & Deprecations

• Class Renames: Renamed long class names to shorter abbreviations for convenience:

  • "mid.importance" -> "midimp"
  • "mid.breakdown" -> "midbrk"
  • "mid.conditional" -> "midcon"

• Argument Updates in interpret():

  • The catchall argument has been renamed to others.
  • The encoding.digits argument has been renamed to digits, and its default value is now NULL.
  • The use.catchall argument is now deprecated and replaced by lump = "none".

3. Bug Fixes & Stability

• Fixed an incorrect calculation in interpret() when singular.ok = TRUE (or weighted.norm = TRUE) and lambda > 0.

• Modified the weights calculation algorithm to enhance numerical stability.