Supervised and Unsupervised Learning

y = f(x)
As described, we have some data points and we want to find a mapping from given data points to some other variable. This is the actual scenario in many machine learning problems, where we are aware of our target variable y and we have sufficient labelled examples to build a machine learning model.

But then there are scenarios when we do not have enough labelled examples, more over some times we are not even sure about the target variable y, we only have some number of data points as x . Here, we wish to make some sense from these data points. 

Former scenario is a typical case of Supervised Learning whereas the latter one is a typical of Unsupervised Learning. We can roughly put Supervised Learning as a task driven approach and Unsupervised Learning as a data driven approach.