Let's talk about data. What does this word mean? Well, it's used in all kinds of context today but I think in a technical practitioner context we mean specifically information which can be used by computers. Now, in the context of machine learning, we are interested in understanding data as composed of columns or what might called variables.

So, data here is always tabular in our model, in our thinking about the problem. So, we have some variables that say x1, x2, x3, and we have a y, and we are thinking in terms of a tabular layout, and what we mean by variable then insert a column or variable. And here we use the letter x to denote a feature or what you might call an observation, an example, a trait, a characteristic, something else. So, here, y we call the target and that is the thing that we are trying to predict. It's a prediction target. Now there are different kinds of connection that y may have to x. In general, we call such connections functions or relationships. And the machine using statistics is able to determine from historical data set and then, by using such a formula, able to, in the future, estimate or predict something for y given what it can see for x.