Module StatsSource

The prbnmcn-stats statistical library.

Introduction

prbnmcn-stats (stats for short) was designed to construct simple probabilistic models and compute basic statistics over empirical data obtained from various experiments.

The core data type is that of a probability distribution. Distributions come under various guises.

• Empirical distributions are the typical outcome of performing a bunch of measurements or experiments, concretely they are nothing more than a collection of data. The empirical mean, empirical variance and quantiles are typical statistics that can be computed on empirical distributions.
• Generative distributions a.k.a. samplers are used when constructing probabilistic models, they are typically used to generate synthetic data.
• Measures are the mathematical representation of a distribution, associating weights to (sets of) data. This is the form we use to compute statistics such as the mean, the variance, etc of a distribution. Typically, we manipulate only finitely supported measures.

stats provides modules dedicated for each case, and maps between the various forms under which probability distributions appear.