Related Posts

Share This

Statistical Learning – Supervised vs Unsupervised

In a supervised learning scenario, we have an outcome, which can be quantitative (real estate price) or categorical (spam email) that we wish to predict based on a set of features (size in square meters, number of rooms). The training dataset procures us both a set of outcomes and features, based on which we will build our model which will allow us to predict outcomes from features that can be new relatively to the training dataset.

In the unsupervised scenario, we do not have an outcome, we only have features. In that case, we will analyze the structure of the training dataset in order to describe trends and clusters. For example, we will cluster customers according to their habits and propose them products they could be interested in.