Comparison criteria strengthened in classification and type representation, according to the Statistical Model of Impact Measuring, in a case study in Pastaza, Ecuador

Resumen

The classifications of five, four, and three groups, obtained using the Statistical Model of Impact Measuring, with the Eta coefficient, were compared, and those corresponding to the confirmatory method of K-Means, which uses the statistical test of Fisher, depending on the estimated impacts of a study of 90 dairy farms. Variable selected were: slope (%), farm area (ha), area used by cattle (ha), compatible area of grazing (%), number of cows (head), female cows for reproduction (head), cows for dairy (head), milk production per year (thousands of liters, mL), number of gullies ha-1 and soil depth (cm). The study was developed in Pastaza province, Ecuador. It was confirmed that for the classifications in five and four groups, 90% of Eta values were higher than 0.60, and for the classification in three groups, 40% were lower. This allowed to conclude that this solution was not the best. The contribution of the three factors obtained with the Statistical Model of Impact Measuring, in the classification of five and four groups, was significant (P<0.001 and P<0.01), according to statistical Fisher (F). Nevertheless, the classification in four groups was selected due to the wide range of factor distribution. The confirmatory analysis of K-Means allowed to improve the classification performed in the Statistical Model of Impact Measuring, because it showed higher heterogeneity among groups and homogeneity within the group, which allowed a better characterization.

Key words: milk production, Statistical Model of Impact Measuring, method of K-means