Clustering of the West African Starchy Roots and Tubers using Nutritive Value

Abstract

The objective of this study was to sub-group the West African starchy roots and tubers’ food sources according to their nutritive values. The k means algorithm was deployed on a dataset consisting of starchy roots and tubers food items extracted from the West African Food Composition Table. Various measures of evaluating clustering validity were employed and the results showed that the clustering was valid. Three clusters were extracted, each consisting of food sources that are very similar in nutritive value. Findings prove that in terms of nutritive value, some kinds of yam and cassava could be classified together, while some kinds of sweet potato and cocoyam could also be classified together, and so on. The clusters can be explored by nutritionists, dieticians, food processing enterprises, and food scientists to find alternative food sources when their original choices are unavailable. Though clustering validation showed that food sources within the same cluster are significantly similar on a general note, a future study is required to research on the extent to which food sources within the same cluster are similar to each other at a granular level.