Producing the diagram

Syntax

BD>pcdiag : , ,

where are the names of bases or elements.

The PCDIAG:  command is used to calculate and store the partial correlation matrix corresponding to a variance-covariance matrix, and to display the calculations on a partial correlation diagram. are assumed to be a list of elements or base names which jointly form the collection . A variance matrix is assumed to have been specified for this collection B, and to be stored currently in the belief store pointed to by the pcsource  control (store 1 by default). The results are available either numerically or in the form of a diagram or both.

Firstly, if the pcdest  control has a value equal to a belief store number, the numerical results detailed above as the matrix G are stored as beliefs in this belief store.

Secondly, if the pcdiag  option is switched on, a partial correlation diagram is output, together with any title defined using the PCTITLE:  command. The diagram consists of the following. Each of the collections is represented as a node on the diagram. The nodes need also to have been given coordinates, colour, and so forth, using the GRID:  or GRID0:  commands. (Any arcs defined within the GRID:  command are ignored.) The nodes are connected by arcs if the partial correlation between the two nodes given the rest exceeds the threshhold given by the pcarc  control, whose value is zero by default.

The arcs are labelled by a small circle showing the degree of partial correlation. We shade the circle according to the absolute magnitude of the correlation, so that correlations of 1 and -1 result in a fully shaded circle. We shade differently, according to direction. For positive correlations, we shade a corresponding proportion of the circle anti-clockwise, starting from 0 degrees. For negative correlations, we shade a corresponding proportion of the circle clockwise, starting from 0 degrees. The size of the circle can be changed using the pcradius  control.