Strategies for analysing nutritional data for epidemiological purposes--food patterns by means of cluster-analytical procedures
1989
Boeing, H. | Klein, U. | Hendrichs, A. | Oltersdorf, U. | Bodenstedt, A.
Results of a study using cluster analysis to identify structures in daily food consumption is presented. The data originate from nutritional records of 86 individuals collected in two seasons. Three strategies were followed when performing the cluster analysis by (1) using different clustering procedures, (2) different lists of food groups, i.e. by including or excluding beverages, and (3) varying the number of clusters. The investigation lead to relatively stable and plausible clusters representing food patterns prevailing in the study population. Of all cluster solutions a 5-cluster partition based on 16 food groups was given priority. According to their consumption profiles the clusters were labeled as "meat", "bread", "fruit", "wine", and "juice" cluster. For this solution the cluster-specific nutrient intake was calculated revealing two well provided clusters compared to the average, two less well provided cluster, and one cluster corresponding more or less to average nutrient intake.
显示更多 [+] 显示较少 [-]