| NEWS | R Documentation |
Now uses JAGS rather than OpenBugs so BTSPAS should run on all platforms. OpenBUGS can still be run using the 'engine' argument on the *_fit() call.
Minor changes to package file structures to meet R package conventions. No change in functionality.
Minor bug fixes discovered in conversion to JAGS.
Initialized missing values in n1, u2, m2 to sensible values to avoid OpenBugs from wandering too far off and generating nonsense values.
Corrected initialization of epsilon when logitP.fixed is set
Minor changes to package file structures to meet R package conventions. No change in functionality.
None.
Plots in non-diagonal case use all time periods rather than just those from releases.
Cleanup of some function calls; Fix arguments for xlim in some plots; don't try and print Rhat if number of chains is 1.
NONE
Fixed and issue where if you had missing values in your data, or zero in n1 and m2 (which the program automatically changes to NA), the scaling factor to compute initial values was returned as NA which caused lm() to fail when initial values are generated.
Posterior Predictive Plots modified to make axes symmetric.
Predictive posterior plots (Bayesian P-value plot) did not have symmetric X/Y axis so the p-value may not print. The axes are now forced to be symmetric.
Misc changes to correct typos, documentation, etc.
BTSPAS now uses OpenBugs exclusively so it can run on Windoze, Mac, and Unix. Users may need to install OpenBugs in place of WinBugs. No other change in functionality.
NONE.