Scientific theories aim to provide a framework for predicting diverse phenomena, and commonly discretize the universe and its behaviour into well-defined entities. Unfortunately, this hierarchical and deterministic approach towards understanding is not always justified, and sharply contrasts with the uncertainty of chaotical behaviour. Consequently, quantum mechanics and relativity have superseded Euclidian space, and the popularity of probabilistic theory has considerably increased. Prediction models have become operating theories of the 21st century, commonly derived from observations now unified as data. The ongoing explosion of data collection, however, also reveals the vulnerability of observation driven research. Particularly, numerous prediction models do not generalize well and perform poorly when applied in new settings. As a consequence, robust estimation and evidence aggregation have become stringent in modern research.