4. Megadata analysis

In this section, we focus on the problem of analyzing very large volumes of data. The nature of this analysis depends on the nature and structure of the megadata, also known as "analytics". These different types of analysis involve various algorithms from data mining, machine learning, decision support and even visualization.

We begin by highlighting the benefits of machine learning for the analysis of megadata. We will then distinguish between the analysis of megadata stored, for example, in NoSQL systems, and the analysis of megadata exchanged and emitted continuously, such as stream data, which cannot be stored due to their volume. As the data concerned by megadata are very diverse in nature and/or level of structuring, their analysis or analysis will be different. We will therefore illustrate a few types of analytics associated with major types of megadata:...