The “Cluster traces using markov clustering ” plugin

TangYahui

 I noticed that the “Cluster traces using markov clustering ” plugin have changed.
The result of the clustering was a number of circles of different sizes, and the size represents the frequency of occurrence.
But now it becomes a circle of the same size with connecting lines between them. Whether the circles connected by lines together mean a cluster？How will I export the cluster now?If I choose them separately, it will be very time consuming。
What I mean is the first picture.

bhompes

Dear Yahui,

Regarding the first part of your question: The tool has not changed. The screenshot tells me you are using the "cluster traces over time" plug-in, rather than the "cluster traces" plug-in. Also, in the two screenshots you have selected different clustering dimensions, so it is to be expected that the results differ as well.

Regarding the second part of your question: I have already answered this question to you before. You can select multiple clusters by dragging an area or by holding the shift or ctrl keys. Note that the mouse mode should be set to 'picking', and not 'transforming'. See your other topic here https://www.win.tue.nl/promforum/discussion/comment/2548.

In the visualization, a circle represents a cluster. A cluster is a set of cases from your event log. Cases are clustered together because they are similar in the dimension you've selected to cluster upon. When two clusters are connected by a line, that means that the cluster share some properties. For example when cases in one of the clusters have similar values for a selection (but not all) of the clustering dimensions compared to cases in the other cluster.

The fact that in your screenshots all clusters are of the same small size means probably that all cases are clustered separately (i.e. no real clustering). This indicates that either the parameters (dimensions, expansion, or inflation) are not set properly or that the event log is not suitable for clustering using these parameter settings.

Kind regards,
Bart

TangYahui

Dear Bart, Thank you very much.Now I can understand them. Kind regards, TangYahui