Στατιστικά μοντέλα ταξινόμησης με εφαρμογή στην ανίχνευση επιπλοκών στη φυσιολογική ανάπτυξη και την καλή υγεία εμβρύων
View/ Open
Keywords
Μοντέλα ταξινόμησηςAbstract
In this thesis, a family of problems known as classification problems and the
techniques for addressing them through statistical classification models are presented.
We focus on three different models that are based on both theoretical and practical levels.
By enhancing them with a set of techniques and methodologies from machine learning,
these models are capable of successfully dealing with problems from various domains
such as finance, medical, data science, and others. The last part of the thesis presents a
real-life problem from field of health sciences, which involves data derived from medical
measurements on embryos through a clinical examination called cardiotocography. The
purpose of the thesis is to apply, compare, evaluate, and optimize the models that were
initially presented at a theoretical level on the medical data, in order to determine which
of them are able to detect effectively any complications in the normal development and
good health of the fetus.