Resumen
The paper offers the methodology of the matrix clustering consisting of multicriteria threshold binarization of the initial matrix of states of objects and clustering the resulting binary matrix into submatrices with different densities of zero and unit elements. Using hand calculation, the methodology was fine-tuned on the export competitiveness indicators of all the Sub-Saharan African countries for the Fresh Food sector of the Trade Competitiveness Map database. A standard R program was developed to implement this methodology and tested for all 14 export sectors of Sub-Saharan Africa, using the data from the Trade Competitiveness Map database for two sets of criteria. It was proposed to automate the procedure of fixing threshold criteria by using the K-Means clustering algorithm for two clusters consisting of zeros and ones.