Changes in version 1.4-2 (2026-05-09) o New perm_test.mmlt. o Update demo reference output. Changes in version 1.4-1 (2026-03-08) o Update to mlt 1.7-4. o Update reference output. Changes in version 1.4-0 (2026-02-26) o All references in inst/REFERENCES.bib. Changes in version 1.3-2 (2025-12-15) o Update Mmlt simulation code. Changes in version 1.3-1 (2025-12-04) o Update survtram vignette. o Update mtram vignette. Changes in version 1.3-0 (2025-11-18) o Adapt to mlt 1.7-0 changes. o Avoid tedious data-driven determination of the support of polynomials in Bernstein form by applying a pre-transformation inspired by nonparametric estimates of the unconditional response distribution. Parameters of the transformation polynomial in Bernstein form are now estimated from input data which is symmetric on the support defined by the range of the response observations. This, specifying support, prob, or log_first in calls to models should be no longer necessary. The old behaviour is still available. o Improve computations defining the support of polynomials in Bernstein form (especially for interval-censored data using the fast Turnbull estimator from package icenReg). o Bring back reference output. o Explain na.action = na.pass behaviour in tram manual page. Changes in version 1.2-4 (2025-09-23) o summary.tram now works for univariate models. o New survtram vignette. o Ship without reference output allowing a more substantial mlt update soon. Changes in version 1.2-3 (2025-06-18) o Update to mlt 1.6-6. o New vignette instroducing nonparanormal marginal adjusted inference (NAMI) as an application of Mmlt. Changes in version 1.2-2 (2025-05-09) o Update survtram demo. o Avoid partial argument matches. o rep to rep_len. Changes in version 1.2-1 (2025-02-25) o Make model.matrix more flexible for tram objects. o New doubly robust score test added as robust_score_test. o Add fit argument to Compris. o Fix broken bib file. o Don't refit instable mixed models for toe nail data in vignette. Changes in version 1.2-0 (2024-11-19) o Adding scores for discrete mtram models. o Update mtram to internal interface changes. o Use mltoptim also in mtram. o Ship proper mtram vignette (so far only the rendered output was supplied). o Make sure demos are in shape (again). Changes in version 1.1-2 (2024-10-04) o Gradient check failed on Fedora noLD. Changes in version 1.1-1 (2024-09-28) o Gradient check failed on MacOS. Changes in version 1.1-0 (2024-09-26) o mmlt moved to mlt 1.6-0. o New Mmlt allowing arbitrary mixes of continuous and discrete data, including exact and censored observations for the same variable; implementing three (potentially) convex approximations; allowing for non-informative missings encoded as -Inf, Inf. o Compris now also uses the parametric likelihood for the competing risk case. o Allow density predictions for all continuous variables (regardless of censoring in training data). Changes in version 1.0-5 (2024-08-17) o Add example for Compris. o Correct simulate.mmlt. o Change to R\_Calloc and friends. Changes in version 1.0-4 (2024-05-13) o Allow larger then needed args$w argument in mmlt. o Fix bug in predict.mmlt. Changes in version 1.0-3 (2024-04-16) o Add sequential fits to mmlt. o Add fixed = TRUE/FALSE argument to coef.mmlt such that the whole parameter vector can be retrieved. o Handle constraints for fixed parameters in mmlt correctly (as mlt always did). o tram and Polr allow binary responses (and return results identical to glm with appropriate family). Changes in version 1.0-2 (2024-01-30) o New demo("survtram") and demo("hcc"). o predict methods for tram and stram now offer the same type argument as mlt::predict.ctm offers. Changes in version 1.0-1 (2023-12-18) o Allow shift-scale transformation models in mmlt. o Add nonparametric log-likelihood to Compris. Changes in version 1.0-0 (2023-08-25) o Improved, simplified, and generalised implementation of mmlt. o Coxph, Colr, and BoxCox allow unbiased estimation of time-to-event distributions under dependent censoring. o All tram models can be equipped with frailty terms. o Fix some minor issues. Changes in version 0.8-2 (2023-03-28) o Update demo stram. Changes in version 0.8-1 (2023-03-09) o Add simulate.mmlt. o Survreg didn't work with constraints, reported by Lucas Kook. o mtram didn't deal properly with interval censoring in all cases. o Enable stratification for shift-scale transformation models (still experimental). o Fix S3 argument issues. o Try to detect negative SEs. Changes in version 0.8-0 (2022-11-02) o Stabilise and generalise mmlt: Alternative link functions, separate lower triangular matrix storage and computations from mmlt. o coef(, type = "Sigma") and coef(, type = "Corr") were incorrect for more than three dimensions. o Improved numerical stability in mtram. Changes in version 0.7-2 (2022-08-07) o New location-scale functionality, see demo("stram"). o New Laplace link function. o New cauchit link function. Changes in version 0.7-1 (2022-05-18) o Minor bug fixed for shift-scale transformation models (still experimental). o bread.tram added for sandwich to work. o Update mtram vignette; move code to demo. Changes in version 0.7-0 (2022-04-01) o Add shift-scale transformation models (experimental). o Add methods for probabilistic indices and overlap coefficients + friends. o Export Gradient and Hessian methods. o Simplify score_test and profile: Use fixed instead of offset. o Models with standardisation now available in mtram for exact continuous outcomes and link functions other than probit. Changes in version 0.6-4 (2022-01-14) o Remove arguments ignored by mlt. o Move mtram vignette code to demo as it took too long on Windows for CRAN. Changes in version 0.6-3 (2022-01-04) o Deal with stratum - covariate interactions, overparameterised coefficients are constraint to zero. Changes in version 0.6-2 (2021-12-07) o Make sure to use the Schur complement when computing variances in for linear transformation models in score_test and perm_test. o Fix LaTeX problems. Changes in version 0.6-1 (2021-09-16) o Add dofit argument to mmlt. o Robustify vcov for mmlt objects. o Add experimental score_test and perm_test methods for glm. Changes in version 0.6-0 (2021-03-10) o Add simulate method. o Enable nonparametric likelihood estimation by Polr. o New argument sparse_nlevels to tram, requesting a model Matrix for ordered factors with at least sparse_nlevels levels. Changes in version 0.5-2 (2020-10-24) o Add experimental score_test and perm_test for survival::coxph. o Update simulation code. Changes in version 0.5-1 (2020-09-04) o Robustify computation of score intervals. o Add Spearman correlation to mmlt example. Changes in version 0.5-0 (2020-08-31) o Lehmann() parameter is now labelled as proportional reverse time hazard ratio. o Add mmlt for multivariate conditional transformation models (continuous responses only for now). o Reproducibility material for added to package. Changes in version 0.4-0 (2020-05-13) o Force support to start at time zero and constrain cumulative hazard to zero for time zero in Aareg. o Add Aareg for Aalen additive hazard models. Changes in version 0.3-3 o score_test generic and method added for computing score tests and corresponding confidence intervals. o perm_test generic and method added for computing permutation score tests. o PI converts log-odds ratios to probabilistic indices and vice versa. Changes in version 0.3-2 (2020-01-12) o profile method added for computing profile likelihood confidence intervals. o Allow additional linear inequality constraints for regression coefficients. o Deal with model_only in Survreg, spotted by Balint Tamasi. Changes in version 0.3-1 (2019-11-07) Bugfixes o Fix problem in C header file. o Fix problems with responses of class R, spotted by Balint Tamasi. Changes in version 0.3-0 (2019-10-22) Bugfixes o y | s1:s2 ~ x lead to rank-deficit models, the intercept is now removed. o Not really a bug, but time-consuming examples were removed (or replaced by static output) in package vignettes to make CRAN happy. o as.mlt returned an object of class mlt but not mlt_fit. New Features o New marginally interpretable models for clustered data in mtram, see mtram vignette. This is still experimental and very rough, so please be careful. o Compute vcov.tram via Schur complement when possible. Changes in version 0.2-6 (2019-06-23) o Add regression test for count data and log_first = TRUE. Changes in version 0.2-5 (2019-02-27) o tram objects are now in line with multcomp::glht, allowing post-hoc multiple comparisons. Changes in version 0.2-4 (2019-01-23) o log_first = TRUE is now accepted by tram and all model functions calling tram. This is still experimental. o Silence regression tests on Windows. Changes in version 0.2-3 (2018-09-18) o New extrapolate = TRUE option for all models constraints the second derivate of the Bernstein transformation function to be zero at support. This helps to get rid of possible discontinuous densities at support. o Add Lehmann for fitting shift transformation models those linear predictor can be interpreted as log-Lehmann-alternative. o Add method = "loglog" to Polr. o New update method. Changes in version 0.2-2 (2018-06-22) o Use default bounds = c(0, Inf) for Surv objects (for example in Coxph. o Update to mlt 1.0-2. Changes in version 0.2-1 (2018-04-23) o Likelihood ratio tests with fixed parameters gave an error. o Remove strata from linear predictor (experimental). o Try to implement a better guess for the add argument later defining the axes corresponding to the response in plots. o All models accept add and bounds arguments. Changes in version 0.2-0 (2018-02-26) o tram first published on CRAN.