Modeling and Predicting

Catalog description

Real life problems are challenging due to their vague description, their multifaceted nature, or their inherent complexity. Dealing with them in a principled way requires model-based thinking. Modeling a problem enables us to abstract, structure, and organize our knowledge about it to make apparent its salient and relevant parts. Thus, models are valuable for multiple reasons. They improve our understanding of a problem, but also highlight the various assumptions that were made to specify them, making explicit their potential limitations. Furthermore, once formulated, models can be tuned and used for prediction (and potentially decision-making later on). In this course, through case studies in various fields (e.g., computer systems, social science, finance), you will learn about various types of models (e.g., parametric or not, causal or not, generative or not…) and their advantages/disadvantages. In addition, for a given problem, you will understand how to select a suitable model, how 001891