Professuren für Statistik und Ökonometrie

Bugs

This page contains a list of (severe) bugs in the different versions of BayesX. The version number of the version installed on your computer can be obtained from the help menu. To report bugs, contact us via bayesx@uni-goettingen.de. Please include any files that are needed to reproduce the bug and describe the problem as precise as possible.

Version: 3.0.2 (17.07.2015)

  • Weights are partially ignored (i.e. set to one, only in equation for mu) for some distribution families with mcmcreg objects. This concerns the families normal, normal2, lognormal, lognormal2, bivprobit and bivnormal.

Version: 3.0.1 (19.02.2015)

  • Estimates of penalized splines obtained from bayesreg objects were not stored correctly.
  • Random slopes could not be estimated with bayesreg objects for non-normal responses.
  • Weights are partially ignored (i.e. set to one, only in equation for mu) for some distribution families with mcmcreg objects. This concerns the families normal, normal2, lognormal, lognormal2, bivprobit and bivnormal.

Version: 3.0 (30.01.2015)

  • Some of the distributional regression methods (multivariate t, dirichlet) were not working correctly.
  • Random intercepts and random slopes could not be estimated with mcmcreg objects and with bayesreg objects for non-normal responses.
  • Weights are partially ignored (i.e. set to one, only in equation for mu) for some distribution families with mcmcreg objects. This concerns the families normal, normal2, lognormal, lognormal2, bivprobit and bivnormal.

Version: 2.1 (07.05.2012)

  • Currently no severe bugs found.

Version 2.01 (21.10.2009)

  • Currently no severe bugs found.

Version 2.00 (6.5.2009)

  • bayesreg objects: BayesX crashes (most of the times) when estimating geosplines. This bug is fixed in version 2.0.1 (21.10.2009)

Version 1.51 (6.05.2008)

  • Currently no severe bugs found.

Version 1.50 (19.04.2007)

  • remlreg objects: BayesX sometimes crashes when estimating time-varying effects in continuous time survival or multi-state models. This bug is fixed in version 1.51 (6.05.2008).

Version 1.40 (18.07.2005)

  • Currently no severe bugs found.

Version 1.40 (06.07.2005)

  • remlreg objects: Offsets are ignored if they are not specified as the last model term. This bug is fixed in version 1.40 (18.07.2005).

Version 1.30

  • remlreg objects: Offsets are ignored if they are not specified as the last model term. This bug is fixed in version 1.40 (18.07.2005).

Version 1.20

  • remlreg objects: Offsets are ignored if they are not specified as the last model term. This bug is fixed in version 1.40 (18.07.2005).

Version 1.11

  • bayesreg objects: DIC is not computed correctly for cumulative probit models. This bug is fixed in version 1.20.
  • remlreg objects: Thresholds in cumulative logit and probit models are not computed correctly (all other results remain valid). This bug is fixed in version 1.20.
  • remlreg objects: Offsets are ignored if they are not specified as the last model term. This bug is fixed in version 1.40 (18.07.2005).

Version 1.1

  • remlreg objects: Results for stationary Gaussian random fields (Kriging terms) are not correct. The bug is fixed in version 1.11 (15.06.2004).
  • remlreg objects: Thresholds in cumulative logit and probit models are not computed correctly (all other results remain valid). The bug is fixed in version 1.20.
  • remlreg objects: Offsets are ignored if they are not specified as the last model term. This bug is fixed in version 1.40 (18.07.2005).

Version 1.04

  • remlreg objects: Offsets are ignored if they are not specified as the last model term. This bug is fixed in version 1.40 (18.07.2005).

Version 1.03

  • bayesreg objects: When estimating a spatial effect with bayesreg objects and using an unordered map with bandsize greater than 40 the estimation procedure does not give correct results. You can
  • check if there are problems using the describe method of map objects. If the bandsize is greater than 40 it is very likely that your estimates are wrong and you should rerun your analysis. This bug is fixed in version 1.04 (20.01.2004).
  • The Pause button does not work properly. This bug is fixed in version 1.04 (20.01.2004).
  • remlreg objects: Offsets are ignored if they are not specified as the last model term. This bug is fixed in version 1.40 (18.07.2005).

Version 1.02

  • bayesreg objects: When estimating a spatial effect with bayesreg objects and using an unordered map with bandsize greater than 40 the estimation procedure does not give correct results. You can
  • check if there are problems using the describe method of map objects. If the bandsize is greater than 40 it is very likely that your estimates are wrong and you should rerun your analysis. This bug is fixed in version 1.04 (20.01.2004).
  • remlreg objects: Offsets are ignored if they are not specified as the last model term. This bug is fixed in version 1.40 (18.07.2005).

Version 1.01

  • bayesreg objects: When estimating a spatial effect with bayesreg objects and using an unordered map with bandsize greater than 40 the estimation procedure does not give correct results. You can check if there are problems using the describe method of map objects. If the bandsize is greater than 40 it is very likely that your estimates are wrong and you should rerun your analysis. This bug is fixed in version 1.04 (20.01.2004).
  • bayesreg objects: Geosplines for not working properly because the variance parameters are not updated. This bug is fixed in Version 1.02.
  • remlreg objects: Offsets are ignored if they are not specified as the last model term. This bug is fixed in version 1.40 (18.07.2005).

Version 1.0

  • bayesreg objects: When estimating a spatial effect with bayesreg objects and using an unordered map with bandsize greater than 40 the estimation procedure does not give correct results. You can check if there are problems using the describe method of map objects. If the bandsize is greater than 40 it is very likely that your estimates are wrong and you should rerun your analysis. This bug is fixed in version 1.04 (20.01.2004).
  • bayesreg objects: Multinomial logit models are not working properly. This bug is fixed in Version 1.01.
  • remlreg objects: Offsets are ignored if they are not specified as the last model term. This bug is fixed in version 1.40 (18.07.2005).

Version 0.9

  • Version 0.9 (11.09.2002): Currently no severe bugs found.
  • Version 0.9 (21.08.2002): Currently no severe bugs found.
  • Version 0.9 (14.08.2002):

    • bayesreg objects: For Gaussian responses and probit models weighted regression does not work properly with P-splines. This bug is fixed in version 0.9 (21.08.2002).

  • Version 0.9 (01.08.2002):

    • bayesreg objects: The program crashes when attempting to estimate random effects (random intercepts). This bug is fixed in version 0.9 (14.08.2002).
    • bayesreg objects: Random slopes are not estimated correctly. This bug is fixed in version 0.9 (14.08.2002).