This research addresses the problem of using data to select the best model from a large space of hybrid models.

For this end, the researcher will develop theory and methods for automated hybrid model selection based on the minimum description length (MDL) principle.

Starting from first-principle models, she will use measured data and machine learning to learn how the parameters of these models vary with other factors. Based on this, the best model can be selected.

As a result, one advantage of this approach is that the researcher can compare different types of models. Moreover, another advantage is that it protects against overfitting without cross-validation, making the approach both efficient and effective.