Applications of Causal Inference

This thesis studies how the field of causality can mathematically define the causal relationships between events distinguishing causal effects from statistically observed correlation. We concentrate on basic causality concepts and definitions of tools and we ask which aspects of the theory can be used to approach practical problems in a systematic manner. This work demonstrates basic causality methods such as building causal models, working with them using d-separation, do-calculus, and methods associated with identification of causal relationships and resolving interventional and contrafactual queries. Necessary assumptions that we need to make before we start building and working with causal models will be outlined. We will demonstrate the theory using simple examples, on the domain of categorical and numerical data. We also will present a simple diagnostic example which aims to fold overviewed tools into a practical application. For all examples data was generated synthetically for the absence of reliable publicly available ground truth data designed for causality. We conclude the thesis by outlining our experience studying and working with the theory, what value we see in using it, as well laying out the benefits and challenges of the theory.