NEWS

spduration 0.17.2.9000

Fix model.matrix so that it works correctly with dots arguments.
Fix an error when trying to use add_duration() or spdur() with a tibble as input. The errors were due to tibble returning lists when subsetting columns with [, col]. The data input is now internally converted to a base data frame (as.data.frame()).
Removed dependency on plyr.

Add pkgdown static documentation website at https://www.andybeger.com/spduration.
Fix C++ code that was causing warnings on R-devel; replacing & and | with && and ||.

Adds proper handling of the various na.action options for spdur and its predict, residuals, and fitted methods.
Import forecast generic function from forecast package rather than re-defining it. Adds forecast to neccessary imports.
Added bug reports link to DESCRIPTION for github issues page (https://github.com/andybega/spduration/issues).
Fixes minor errors in introduction vignette and data documentation.
Register C++ routines to avoid R-devel note.

separationplot.spdur -> sepplot. separationplot is a standalone function so change the spdur version to a simple wrapper.
Cleaned up NAMESPACE. Moved stats to imported package in description and removed all associated explicit namespace generic imports.
Added package vignette.
Added file init.c with calls to R_registerRoutines() and R_useDynamicSymbols(); also use .registration=TRUE in useDynLib in NAMESPACE. R-devel (and R 3.4.0 in the future) R CMD check issues a NOTE for registration of routines, this is to avoid that note.
Updated DESCRIPTION and added references.

Fixes bug in summary.spdur that would return wrong estimates for log(alpha).
Better column names for summary and xtable methods.
Fixes test error after testthat update.

Added accessor methods for terms, model.matrix, coef, and vcov.
Added fitted and residuals methods.
Fixes issue #17, which led to errors in summary.spdur when called on a model with factor variables or without intercept terms.
Fixes an issue with the hazard rate plot confidence intervals (plot_hazard(ci = TRUE)), where CIs coud be wrong because coefficients were sampled by equation rather than using the full variance covariance matrix.

Added Belkin & Schofer 2003 coup data in data(bscoup), see ?bscoup.
Replaced plot_hazard1 and plot_hazard2 with plot_hazard, added support for loglog models, and other internal streamlining of plotting code.
Partial fix for summarizing and printing results for models without intercept term.
Removed documentation and export for several internal functions.