How to do classification using clustering? Is there any open source for the same.
It depends on your vision of your data. Clustering is generally considered a "non-supervised" technique in Artificial Intelligence terms, whereas classification is "supervised". "supervised" means that we have some label which tells us the class of each record (for example, "mammal" or "non-mammal"). If we don't have the label, we can cluster the data using the available attributes, and then inspect each cluster to see its statistics in terms of the attributes. If the attribute-values in cluster1 are mainly characteristic of mammals and the attribute-values in cluster2 are mainly characteristic of non-mammals, then we can say that the clusters have "classified" the records with respect to the "mammal" criteria. You can use a standard k-means clustering algorithm to do this.
As par as my concern classification and clustering both are using to make groups of given data set.
But classification do the things in supervised manner , like well known about data according make data groups. Other end clustering is unsupervised means we do not know how many groups in the data.
Example :
In the class room 60 students marks are there
In the supervised manner (classification) saying that group the 60 students into three according to their marks like First class , Second class, Third class.
Other way in unsupervised (clustering) we do know how many groups to do, even what each group contain or behave.
This is the basic difference between classification and clustering.
But in your question u asked how to do classification using clustering Make your question little clarity please
- Badges
- Science topic
- Similar topics
- Computer Science
- Data Mining