NEWS

---

coxed 0.3.7

• fixed a bug with sim.survdata() in which user-supplied X matrices are replaced with randomly generated X matrices with type='tvbeta'

coxed 0.3.6

• fixed a bug that occurred with sim.survdata() with type=‘tvbeta’, user-specified coefficients, and one covariate. Collapsed the separate code for the one covariate case into the more than one covariate case by coercing beta into a matrix so that ncol() works.

coxed 0.3.4

• changed the method through which the sim.survdata() function warns users that their linear predictor is too large to produce good results, and should reduce the size of the beta coefficients, and changed the text of the warning

coxed 0.3.3 (2020-08-02)

• changed the call to rms::cph() to work with recent changes to the rms package

coxed 0.3.2 (2020-01-10)

• fixed a bug that prevented the vignettes from being compiled on windows machines

coxed 0.3.1

• fixed a bug that led to (very close to 0) negative values of the survivor function when the random spline failure CDF reaches 1 prior to the final time point

coxed 0.3.0 (2019-08-01)

• edits to help documentation and vignettes in preparation for release to CRAN

coxed 0.2.7

• fixed bug with coxed.npsf.tvc() in which all prior time points for every observation were included in the risk set. Changed it so that all yet-to-fail observations are included using ONLY the time point in question

coxed 0.2.6

• rewrote rank.predict() to make GAM much faster with out of sample prediction with GAM

coxed 0.2.5

• fixed a bug in cox.gam.tvc.R and cox.npsf.tvc.R in which the functions failed to produce out of sample predictions

coxed 0.2.4 (2019-04-12)

• fixed typos in vignettes

coxed 0.2.3

• fixed a bug with sim.survdata() with type=“tvc” that prevented properly setting the amount of right-censoring

coxed 0.2.2

• fixed a bug with the TVC routines. All observations were being used to calculate cumulative baseline hazard for the NPSF function, resulting in a larger risk set, leading to lower hazard and longer durations than are correct. We fixed the bug by calculating the CBH from only the last observation of each unique value of ID

coxed 0.2.1

• edited and rebuilt vignettes 8/24/2018

coxed 0.2.0 (2018-08-23)

• updated version number to reflect the added feature, and fixed minor bug 8/23/2018

coxed 0.1.3

• adds bias-corrected and accelerated (DiCiccio and Efron 1996) confidence intervals for bootstrapped samples in addition to the studentized and empirical options 8/23/2018

coxed 0.1.2

• changes a sentence in the coxed.Rmd vignette, and updated the version number for the next update 6/15/2018 # coxed 0.1.1

• first working version, ready for extensions/debugging/meeting CRAN demands 6/13/2018

• changes for CRAN resubmission: reduced the coxed() example to run in half the time, and reduced the computational requirements of the vignettes 6/14/2018

coxed 0.1.0

• initial release 6/13/2018