next up previous
Next: Adjusted expectations Up: Adjusting beliefs by data Previous: Adjusting beliefs by data

Performing an adjustment

 

In our first example we obtain the adjustment of the belief structure tex2html_wrap_inline9695 by belief structure tex2html_wrap_inline9697 . Here is the [B/D] command which determines the adjustment (remember that the quantities tex2html_wrap_inline8606 and tex2html_wrap_inline8608 in the command are bases which we have already definedgif ):

BD>adjust:[B/D] tex2html_wrap_inline6116

No output resulted when you typed this command (unless you made a mistake, in which case you may re-enter the command). This is because [B/D] only reports results that we explicitly request, and whilst we could have requested various results beforehand (but didn't), we will find it more convenient here to use the SHOW:  command as in the following sections.

The ADJUST:  command is of key importance within the program, and there are a great many program switches which can influence what happens precisely when you issue such a command. For our purposes, the default configurationgif is suitable.

The variance and covariance specifications made in section 3.2 are coherent in that the variance-covariance matrix for the vector tex2html_wrap_inline9703 is non-negative definite. Such non-negative definiteness is a requirement for our approach (that is, we disallow the possibility of negative variances). The ADJUST:  command automatically checks that your various inputs are coherent, and advises you when they are not, by reporting the error and halting the adjustment.

Many of the effects of an ADJUST:  command can be viewed by supplying various options to the SHOW:  command. An option in [B/D] is used to specify the type and amount of output that we see. For many options, greater detail can be had by changing the pertinent output option slightly. In fact we add a `+' symbol to the option. To help you, we will show the names of options in a different typeface like this . Initially we look at three of the most basic types of output: adjusted expectations and evaluations of them, and adjusted variances.


next up previous
Next: Adjusted expectations Up: Adjusting beliefs by data Previous: Adjusting beliefs by data

David Wooff
Thu Oct 15 12:20:04 BST 1998