help sqclusterdat                                               (SJ6-4: st0111)


sqclusterdat -- Add solution of MDS on dissimilarity matrix to sequence data


sqclusterdat [, return keep]


If the command sqom was performed with option full, further analysis of the resulting dissimilarity matrix becomes indispensable. Cluster analysis is the most common technique for this step. Unfortunately, sequence data and the resulting dissimilarity matrices have different dimensions, so that the cluster variables cannot easily be attached to the sequence data on a row-by-row basis. sqclusterdat allows linking the results of user-specified cluster commands to the original sequence data.

Without return, the command constructs a dataset that is built from instructions left over from the last sqom command. The user may then specify arbitrary clustermat commands, as well as applicable postclustering commands in this dataset. After performing the cluster analysis, sqclusterdat, return brings the cluster results back into the original sequence data.

Note that sqclusterdat, return only works if you have specified the option add in for any cluster analyisis performed with clustermat.

The option keep(varlist) is usable for clusterdat, return only. It is used to specify a list of hand-made variables that should be merged to the original dataset as well. This is especially useful for variables created with cluster generate.

The convenience command sqclustermat allows performing the three steps with one single command. Cluster postestimation commands do not work in this case, however.


Ulrich Kohler, WZB,


. sqom, full k(2) . sqclusterdat . clustermat singlelinkage SQdist, add name(mydist1) . clustermat singlelinkage SQdist, add name(mydist2) . cluster tree mydist1 . cluster generate gr = groups(2) . sqclusterdat, return keep(gr)

Also see

Manual: [MV] clustermat

Online: sq, sqdemo, sqset, sqdes, sqegen, sqstat, sqindexplot, sqparcoord, sqom, sqclusterdat, sqclustermat