Once suitable data is available, you may now want to start building models of the data. Models are usually built to explain significant relationships in the data or to predict future states based on the past data. The models required and the way they are tested will depend on the type of data and the goal of the modelling.

Modelling can take the form of a simple statistical model, a computationally intensive deep neural network, or anything in-between. The choice of modelling technique will depend on the problem to be solved and the resources available but in all cases the most simple, interpretable model that can be used, should be used.