| Description |
Welcome to this applet, demonstrating the method of Clustering by a little
toy example.
Please note, that Clustering is a useful tool for serios applications, like
dividing a group of customers in similar subgroups, so they can better be
addressed by advertising, etc.
The setting of this example is a party, where different persons meet and
talk to each other. Let's assume, that all guests prefer to talk to someone
with similar interests, etc. In the applet people are described by six different
properties, which are:
- if they have children,
- if they are married,
- which drinks they prefer, this evening,
- their hobbies,
- their favorite games,
- and which car they are driving.
Please don't take this list to serious!
Depending on the given properties, the clustering algorithm COBWEB
builds groups of similar persons. The node on the top represents the group
of all party guests. If you follow the edges downwards, you reach smaller and
smaller subgroups of more and more similar people, until you reach a leaf.
Each Leaf represents a single guest, which, of course is perfectly similar to
him- or herself.
Inviting Guests:
On the upper part of the applet (gray background) you can invite new guests to
your party. Just choose the properties of your new guest, using the choice-boxes,
and click on the "Invite to the Party!"-Button, when you have
finished. Clicking "Undo!" will bring you back, to before the
last invitation. You can also invite a group of seven "Default
Guests", by clicking on the corresponding button.
These guests are added one by one, with a certain delay.
To see, what a node in the graph represents, please click on the node, and
refer to the text field below the graph.
Have fun!
|
