So far we have worked with probabilities known a priori. Descriptive statistics deals with the inverse problem: estimating quantities (means, proportions, parameters) from an observed sample. It is the bridge between the mathematics of probability and the world of real data: surveys, scientific experiments, diagnostic tests, quality control. Before computing the indices it is worth looking at the data with a good graphical representation, which immediately shows the shape of a distribution (symmetric, asymmetric, multimodal, with outliers) where a single number might hide it.