Possibilistic Linear Regression Analysis with Fuzzy Response Variable for Crop YIELD Estimation

Abstract

Reliable estimates of crop yields at small area level, say blocks, are of great importance for policy planning at micro-level. To this end, application of present methodology of Crop-cutting experiments is not practicable, as it would require total number of such experiments to increase many folds. Additional information about farmers' estimates of crop yields at block level, which are crisp values, may be used provided these can explain the actual crop yields, which are fuzzy. Accordingly, in this paper, theory of fuzzy sets and possibilistic regression analysis is discussed. Three methods, viz . Minimization, Maximization, and Conjunction are considered. The methodology is applied for modelling cotton crop yields at block levels of Sirsa district, Haryana. It is found that Conjunction method performed the best. Further, farmers' estimates are able to explain the actual crop yields with fitness level as high as 0.6.