В Мичоакане создан индикатор конкурентоспособности для производителей голубики

A doctoral research project conducted in Michoacán has developed an Agricultural Competitiveness Indicator (ICA) for blueberry growers, combining statistical analysis and machine learning to assess business capabilities and external conditions associated with competitiveness. The research considers factors beyond output, yield, exports, prices, and cultivated area. These include skilled personnel, access to financing, technology, cost management, water availability, and marketing channels. Based on literature and information from participants linked to the blueberry sector, researchers developed a quantitative instrument containing 42 items, which was subsequently applied to growers. The questionnaire recorded a Cronbach's alpha of 0.947 and a KMO index of 0.875, while Bartlett's test was statistically significant. Exploratory factor analysis (EFA) was then used to identify seven competitiveness factors, explaining 78.69 per cent of the variance. The seven dimensions were Strategic Capabilities, Markets and Logistics, Governance, Production, Financing, Bureaucracy, and Water. The results indicate that agricultural competitiveness is influenced by both company capabilities and the operating environment. Random Forest machine learning was subsequently used to assess which factors provided the most information for differentiating competitiveness levels. Strategic Capabilities recorded the highest predictive importance at 31.27 per cent, followed by Financing at 21.06 per cent, Production at 17.9 per cent, and Water at 12.61 per cent. Markets and Logistics accounted for 8.98 per cent, Governance 5.98 per cent, and Bureaucracy 2.2 per cent. Some factors produced similar results under both methods. Strategic Capabilities represented 32.08 per cent in the EFA and 31.27 per cent predictive importance under Random Forest. Water showed a larger difference between the two approaches. It represented 4.52 per cent in the EFA but reached 12.61 per cent predictive importance under Random Forest. According to the research, EFA identifies the statistical structure of the factors, while Random Forest provides a separate assessment of their predictive importance within the model.