Why use classification at all after clustering algorithm in Machine Learning?

Question

Hi, I am learning Microsoft Azure AI Fundamentals: AI Overview and I have a question regarding a scenario given in Types of machine learning module which says:

In some cases, clustering is used to determine the set of classes that exist before training a classification model. For example, you might use clustering to segment your customers into groups, and then analyze those groups to identify and categorize different classes of customer (high value - low volume, frequent small purchaser, and so on). You could then use your categorizations to label the observations in your clustering results and use the labeled data to train a classification model that predicts to which customer category a new customer might belong.

Why we need to use a Classification model to predict which customer category a new customer might belong to? Can't I directly feed the new customer to a clustering model which will segment it into respective group(which it has already done for my past customers data as given in above paragraph)?

This question is related to the following Learning Module

Answer 1

Hi Chetan Raut,

We utilize a classification model to predict the category of new customers because clustering, as an unsupervised learning technique, does not preserve grouping logic in a form that can be directly applied to new data. Clustering is primarily used in the initial phase to identify natural patterns and groupings within the existing dataset. However, incorporating a new customer into a clustering model typically requires re-executing the entire clustering process, which can be computationally intensive and may alter the existing group structure.

Once clusters are established and interpreted (e.g., labeled as "high value – low volume" or "frequent small purchaser"), a classification model can be trained using this labeled data. This model enables consistent and efficient prediction of customer categories for new data points, without impacting the original cluster definitions.

In this way, classification offers a practical, scalable, and reliable solution for applying the insights gained from clustering to future customer data.

Please reach out to us if you have any other queries.

If the information is helpful, please Accept Answer & Upvote so that it would be helpful to other community members.