Τεχνικές ομαδοποίησης μικτών δεδομένων
Mixed data clustering techniques
View/ Open
Keywords
Clustering ; Mixed data ; Μικτά δεδομένα ; Ανάλυση σε ομάδες ; k-means ; KAMILA ; DBSCAN ; Agglomerative hierarchical clusteringAbstract
In Statistics, clustering analysis techniques aim to create homogeneous groups, while at the same time ensuring that groups are as different as possible from each other. Consequently, by examining each group separately we can achieve easier and more efficient processing for the available data. Mixed data sets, i.e., data that include both numerical and categorical characteristics, arise in various fields such as healthcare, finance, marketing, and others. Typically, the application of cluster Analysis techniques to mixed data sets is exploited in order to identify structures and group similar objects for further analysis. The aim of this thesis is to present and apply mixed data clustering techniques to simulated data and perform statistical analyses on real data. An evaluation of these techniques will also be performed using the results obtained from the applied techniques on simulated data, as well as through a bibliographic review of these techniques on real data.