Identification of homogeneous geographical areas of mortality for tumours from cluster analysis

This paper attempts to demonstrate the utility of cluster analysis as a descriptive method of studying mortality in epidemiology. In order to verify which algorithms of clustering best fit the data structure, the method of cophenetic correlation was implemented. Furthermore the probabilistic algorithm proposed by Beale was used to assess the partition. The results show the presence of some striking clusters between Local Sanitary Units of the Emilia Romagna Region for four types of tumour in men.