Νευρωνικά δίκτυα και μηχανική μάθηση, εφαρμογές στη βιοϊατρική και βιοπληροφορική
Neural networks and machine learning, applications in biomedicine and bioinformatics
View/
Keywords
Μηχανική μάθηση ; Αλγόριθμοι ταξινόμησης ; Λογιστική Παλινδρόμηση (LR) ; Μηχανές Υποστήριξης Διανυσμάτων (SVM) ; Αφελής Μπεϋζιανός (NB) ; Μετρικές απόδοσης ; Πίνακας σύγχυσης ; Καμπύλη ROCAbstract
In this thesis, the problem of predicting the occurrence of heart attack based on symptoms using Machine Learning algorithms for binary classification is investigated in order to compare the results of these algorithms. The use of predictive methods with the help of data mining and especially Machine Learning can contribute to early diagnosis to avoid serious health complications in patients. The implementation was done with the help of the Scikit-learn library of the Python programming language with a specific dataset publicly available from the Kaggle repository. The dataset is observations collected from 918 patients, which includes the most common symptoms of the disease. The machine learning algorithms applied to the dataset are Logistic Regression, Support Vector Machines and Naive Bayes. The results were compared with various performance metrics related to classification problems. The comparative results indicate Naïve Bayes as the best algorithm for the given dataset. The associated model shows the best performance metrics, with Support Vector Machines following.
Postgraduate Studies Programme
Ψηφιακός Πολιτισμός, Έξυπνες Πόλεις, IoT και Προηγμένες Ψηφιακές ΤεχνολογίεςDepartment
Σχολή Τεχνολογιών Πληροφορικής και Επικοινωνιών. Τμήμα ΠληροφορικήςNumber of pages
67Language
GreekCollections
Except where otherwise noted, this item's license is described as
Αναφορά Δημιουργού-Μη Εμπορική Χρήση-Όχι Παράγωγα Έργα 3.0 Ελλάδα
Αναφορά Δημιουργού-Μη Εμπορική Χρήση-Όχι Παράγωγα Έργα 3.0 Ελλάδα