Visualization Tab

Select the Visualization tab of the Link Analysis Results dialog box to access the options described here.

Web graph (support & lift)

Click this button to display a graphical summary for the final set of the association rules for the specific items selected in the Item name list view (located in the Container of the Link Analysis Results dialog). In this graph, the support values for the body and head portions of each association rule are indicated by the size of each circle. The thickness of each line indicates the relative joint support of two items and its color indicates their relative lift. A minimum of two items in the Item name list view must be selected in order to produce a web graph.

Rule graph (support & confidence)

Click this button to display a graphical summary for the final set of the sequence/association rules for the specific items selected in the Item name list view (located in the Container of the Link Analysis Results dialog). In this graph, the support values for the body and head portions of each rule are indicated by the size of each circle. The colors of the circles indicate the relative confidence. A minimum of two items in the Item name list view must be selected in order to produce a rule graph.

With all single items

Clicking a results button (spreadsheet or graph) contained in the With all single items group box (Sequence rules, Frequent itemsets, Support graph, and Confidence graph, described below) will produce results for all items in the Item name list. Clicking the other results buttons will produce results only on the selected items in the Item name list.

Clustering of items

Displays the graph of single linkage clusters.

Clustering of items

Click this button to display a spreadsheet showing the cluster of the frequent itemsets. Specifically, the spreadsheet will report, for each itemset, the number of items, frequency and support value.

Disjoint sequences

Click the Disjoint sequences button to display a spreadsheet of sequences in which none of the clusters overlap. The subsets shown in this spreadsheet will not have any items in common. For example, consider a set of sequence events defined by the data set {a, b, c, d, e, f, g} and for which the resulting set is S = {a:4, b:3, d:5, f:10}. We can define a subset that contains the last two items (d:5 and f:10), and another subset that only contains the last item (f:10). These two subsets are not disjoint, because they have overlapping elements (f:10 is in both subsets). A collection of disjoint sequences for this simple example could be a set that only contains a:4, a set that only contains b:3, and a set that contains d:5 and f:10. The rules for disjoint sequences are the most complex rules that can be extracted from a dataset.