Συγκριτική αξιολόγηση αλγορίθμων μηχανικής μάθησης σε δεδομένα ασθενών με διαβήτη
A comparative evaluation of machine learning algorithms in patient data with diabetes
View/
Keywords
Κ κοντινότεροι γείτονες ; Μηχανική μάθηση ; Απλοϊκός Bayes ; Λογιστική παλινδρόμηση ; Νευρωνικά δίκτυα ; Μηχανές διανυσμάτων στήριξης ; Δένδρα απόφασης ; Συλλογική μάθηση ; Σακχαρώδης διαβήτηςAbstract
The rapid development of technology in recent years, coupled with the daily storage of large volumes of data, and the need for a proper classification of these data, has led to the implementation of various algorithms for their better classification. Classification takes place in many scientific fields such as medicine, economics, meteorology and much more. In the field of medicine, the correct and immediate prediction of a metabolic disease such as diabetes plays a very important role. These diseases can cause several other more serious complications, so it is conceivable that developing systems where they can predict these diseases with big accuracy is very important.
This dissertation titled "A Comparative Evaluation of Machine Learning Algorithms in Patient Data with Diabetes" refers with the comparative evaluation of the performance of mechanical learning algorithms and in particular supervised learning for predicting diabetes. Such algorithms are Bayesian simplistic, logistic regression, neural networks, support vector machines, decision trees, collective learning and the “K closest neighbors”. A summary of the different algorithms selected, the results of other international studies on the same topic, are presented, and then the results of applying the algorithms to two sets of data, available for free study online. The editing was done using the Python programming language.
Postgraduate Studies Programme
Ψηφιακά Συστήματα και ΥπηρεσίεςDepartment
Σχολή Τεχνολογιών Πληροφορικής και Επικοινωνιών. Τμήμα Ψηφιακών ΣυστημάτωνNumber of pages
132Language
GreekCollections
Except where otherwise noted, this item's license is described as
Attribution-NoDerivatives 4.0 Διεθνές
Attribution-NoDerivatives 4.0 Διεθνές