NEWS

---

modeldiag 0.1.2

• Added VIF severity levels and per-predictor reporting:
  • Severity mapping: < 2 = Negligible; 2–5 = Moderate; 5–10 = High; >= 10 = Severe.
  • check_vif() now returns severities and collinear_predictors (predictors with VIF >= 10).
  • summary() prints per-predictor severities and a severity legend.
  • Tests updated to cover severity labels and collinear predictor listing.
• Added linearity diagnostics for linear models using lmtest::resettest().
  • diagnose_model.lm() now includes linearity in its test suite.
  • The summary output interprets whether there is evidence against linearity.

modeldiag 0.1.1

• Improved multicollinearity reporting in summary() for linear model diagnostics.
• Fixed Authors@R metadata so package builds derive Author and Maintainer correctly.

modeldiag 0.1.0 (2026-05-28)

• Initial CRAN submission.
• Added diagnose_model() generic function with methods for:
  • Linear models (lm)
  • Generalized linear models (glm) with binomial and poisson families
  • Cox proportional hazards models (coxph)
• Implemented comprehensive diagnostic tests:
  • Multicollinearity (VIF)
  • Heteroscedasticity (Breusch-Pagan test)
  • Autocorrelation (Durbin-Watson test)
  • Normality (Shapiro-Wilk test)
  • Influential observations (Cook's distance, dfbetas)
  • Overdispersion (Poisson models)
  • Zero-inflation (Poisson models)
  • Proportional hazards (Cox models)
  • Linearity of logit (Logistic models)
  • Goodness of fit (Hosmer-Lemeshow test)
• Added print(), summary(), and plot() methods for diagnostic results
• Comprehensive plotting capabilities for visual diagnostics
• Full documentation with roxygen2
• Unit tests with testthat