If you find a bug, or if you have any comments/criticisms/questions about this package, then please feel free to email the package developer.

OPEN ISSUES


BUGS

None known.


FEATURES TO BE IMPLEMENTED

Large-scale changes are planned for version 1, including removal of the bayescount and likelihood/maximise.likelihood functions.  I hope to have these submitted to CRAN in mid-2015.  Additional options for precision, power and analysis of count and FECRT models will be available in a more coherent manner



VERSION HISTORY

Version 0.9.99

NEW FEATURES

A more robust fecrt.model function with more options

Functions to be removed from version 1 are now deprecated

BUG FIXES

Issue fixed preventing the citation file being parsed correctly by CRAN

New contact details and bug fixes in manuals



Version 0.9.9

NEW FEATURES

Help file concerning the likelihoods produced by bayescount and bayescount.single updated for clarity.

power analysis

fecrt function now takes confidence parameter - used for all CI intervals and boot/mcmc classifications (from probability)

fecrt function and all gamma models in bayescount now use a T(10^-200,) on the gamma distributions to prevent crashes caused by a failure to calculate log densities associated with very small gamma values


BUG FIXES

Bug fixed which could cause a crash with bayescount when using the scale mean option with data consisting of all zeros

Bug fixed which would prevent bayescount from running with data that could have been multiplied by a constant

Citation information updated for compatibility with R version 2.10

Numerous other minor bug fixes


Version 0.9.0

NEW FEATURES

Functions for running JAGS developed into 'runjags' package, which is now a requirement for bayescount.

Running of models to convergence and calculation of necessary run time is now handled using the autorun.jags function.

NAMESPACE file used now

Model specification for (ZI)GP and LP models slightly altered.  The new model specifications produce superior results.

bayescount.single now calculates 95% credible intervals using highest posterior density estimates rather than quantiles

Likelihood calculations now also return the maximum likelihood estimate

Repeat counts can now be analysed (although the likelihood cannot be calculated with irregular numbers of counts or with any repeat counts for the Weibull models)

Coefficient of variation can now be specified as a parameter to lnormal.params and normal.params, and functions now return lists rather than a matrix.

FECRT function now implemented.  See ?fecrt

Several other new features


BUG FIXES

Superficial changes to the feedback provided by the likelihood function.

Superficial changes to wording of feedback provided by bayescount and bayescount.single.

Several other bug fixes.


OLDER VERSIONS

Function to call JAGS for any given model/data/inits with any number of chains created November 2007.

Modified to use JAGS 0.9.99 and lognormal / Weibul / indpendant Poisson models November 2007.  Also modified to extend rather than re-run for extra convergence iterations.

Function to analyse faecal egg count data with either a (zero-inflated) Gamma Poisson, Lognormal Poisson, Weibul Poisson, independant Poisson or single Poisson model then check converegence with 2 chains created October 2007.

Source script to handle/call function to analyse faecal egg count data created 3rd July 2007.