Search code examples
wekadata-mining

Why do I need classify Data after clustering?

I'm almost new in Data Mining. I Clustered a set of data. Why do I need classify? For example my clustered data is as:

cluster A:
Age: 25
average: 12.5


cluster B:
age: 24
average = 14

that I clustered data for predict 'sex':

A --> Male
B --> Female

First can I conclude that Female has a greater average? if so, Why i do need use classify?

Solution

  • You do not need to classify after clustering.

    Your approach does not make sense to me, and your result most likely is incorrect.

    If you want to compute the average ages, compute them on the M/F subsets of your data. Do not assume your clusters agree to e.g. gender this will usually not work. If you know your desired clusters, don't use clustering but use your classes.

    Clustering always adds uncertainty to you result. Only use it to explore structure in your data, but never assume a clustering to be correct. There is no such thing. It is a tool to help you find new ways of looking at your data, but it must no replace critical thinking and human inspection.