Εφαρμογή τεχνικών ελάττωσης διάστασης πολυδιάστατων δεδομένων στη δημογραφία
Application of dimensionality reduction techniques in demography
KeywordsΜεταβλητές (Μαθηματικά) ; Δημογραφία ; Παραγοντική ανάλυση αντιστοιχιών ; Ανάλυση δεδομένων ; Στατιστικές μέθοδοι
The main purpose of the below presented dissertation is to graphically highlight the correlations that may exist in a number of categorical variables. More specifically, we have in our disposal demographic and health data for older people over 50 years of age in different European countries in 2013. The method that will be used is called Multiple Correspondence Analysis and is the generalization of Factorial Correspondence Analysis used for the analysis of only two categorical variables. Through this study, an association was emerged between groups of people with specific demographic characteristics and the presence or absence of specific diseases. The first chapter aims to inform the reader about the founders of the method and its evolution over time. The second chapter presents the theoretical background of the method, initially for two variables and then generalizes in the multivariate case. In the third chapter we present our data set, while in the fourth chapter we apply the method by testing various models while presenting and comparing the results obtained. Finally, in the fifth chapter we present the general conclusions regarding the structure of the data and the correlations discovered by the various analyses.