Clustering analysis is an unguided analysis. If you understand the core meaning of clustering, you can understand that the number of clusters is not standard and must be set artificially, but special clustering methods can give you some reference, such as systematic clustering, which can generate clustering trees, so that you can intuitively judge whether it is appropriate to divide into several categories. Another example: second-order clustering, the system model will automatically choose to be divided into several categories (if not artificially set).

clustering is an unsupervised training sample, and classification is a guided training sample. Classification is to establish classification rules through known samples to predict the classification of new samples. Why is it guided? Because classification is a model that uses other attributes of samples to explain and predict the attributes we are interested in, this is vernacular. For example, we know the demographic variables, consumption, wages and loan repayment of a group of users. Now we need to use the demographic variables, consumption and wages of users to predict the loan repayment of users. This is the classification model, and we need to use the classification decision tree here. That is to say, we use other attributes of the sample to establish classification rules for the loan repayment of the sample, and then predict the new sample in the future to judge whether the new user is an ideal lending object.