Abstract:

Events in nature occur with the effect of many interrelated variables, either separately or together. It is important to introduce and use the methods used for the analysis of data sets at different scale levels with linear and nonlinear relationship structure between variables. Redundacy Analysis and Canonical Correspondence Analysis are among the methods used in the analysis of such data. Aforementioned techniques are generally carried out by ecologists and there are limited studies in the field of health. In the study, the application of the methods was performed with a data set in the field of Cardiology including variables at different scale levels and their performances were compared. Determination Coefficient (R2) and MAPE (Mean Absolute Percentage Error) value were calculated as performance criteria. According to the results, it was seen that CCA and RDA, which analyze the relationship structures between variables in different scale types (cardiological data set), explain the variation sufficiently. Also, it was emphasized that both methods classify well with low MAPE value (less than 10%) and perform ordination diagram. In addition, it has been observed that restricted ordination diagram models give satisfactory results in determining the relationships between coronary heart disease data and so that they can be used in the field of health too.

Keywords: