What are the fundamental differences between predictive modeling and clustering? according to literature, predictive modeling is a supervised learning with aim to construct models to predict the value of target attribute. On the other hand, clustering is a unsupervised method to split a data set to a couple of groups. In my idea, both of them are the same. when you are finding out the target attribute, you are actually trying to categorize it which is similar to clustering. I know that in clustering, we make subgroups of available data while in predictive modeling we try to find a model to predict value of attribute of unseen data based on current available data. Could anyone explain the differences between these two concept clearly? Thanks
Predictive models are sometimes called learning with a teacher, whereas in clustering you're left completely alone.
Predictive models split data into training and testing subsample which is used for verifying computed model. Predictive (or regression) model typically assign weights to each attribute. From clustering you can compute some internal evaluation metric, but that doesn't necessarily correlate with (desired) human judgment. Unsupervised learning (clustering) mostly treats all attributes as equal as without external information, as one can't say which attribute is more important than the other.