25.12.2013 Views

CRANFIELD UNIVERSITY Eleni Anthippi Chatzimichali ...

CRANFIELD UNIVERSITY Eleni Anthippi Chatzimichali ...

CRANFIELD UNIVERSITY Eleni Anthippi Chatzimichali ...

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

Linkage algorithms consist of various different algorithms such as single linkage,<br />

complete linkage and Ward’s method. For two given clusters and , a single<br />

linkage algorithm calculates the shortest distance between the two clusters as<br />

described in Equation 5<br />

‖<br />

‖<br />

Equation 5 Single linkage algorithm<br />

Where and are elements of the clusters and respectively.<br />

1.4.2.2 -means Clustering<br />

-means clustering is an unsupervised clustering technique that attempts to “minimise<br />

the sum of point-to-centroid distances, summed over all clusters” (Arthur and<br />

Vassilvitskii, 2007). The objective function for -means clustering is<br />

∑ ∑‖<br />

‖<br />

Equation 6 k-means clustering algorithm<br />

Where, is the number of clusters, is the th pattern belonging to the<br />

th cluster,<br />

is the centre of cluster and is the number of data points. The algorithm’s steps<br />

can be described as follows:<br />

1. Initially, the number of clusters is selected, for instance<br />

2. Randomly the items are assigned to the clusters<br />

3. A new centroid is calculated for each of the clusters (a distinct set of points<br />

belongs to a certain centroid)<br />

4. The distance of each item towards the centroids is calculated<br />

5. The items are subsequently assigned to the closest centroid<br />

6. The algorithm keeps iterating until the assignments to the clusters are stable<br />

The algorithm’s simplicity and speed makes it an appealing technique for cluster<br />

analysis. However, a major disadvantage of this algorithm is that it is sensitive to the<br />

selection of the initial partitions.<br />

13

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!