A goal programming approach for developing pre-harvest forecasts of crop yield

In this paper, an attempt has been made to develop a pre-harvest forecast of sugarcane yield. The forecast is based on plant biometrical characteristics such as plant height, girth of cane, number of canes per plot and width of third leaf from the top. Some of these characteristics are correlated and hence the conventional practice of fitting a regression model using least square technique for estimating the parameters does not lead to satisfactory pre-harvest forecasts. Keeping in view the difficulties relating to the violation of the assumptions of normality, independence and homoscedasticity of classical multivariate regression analysis, the present study proposes an alternative approach, which is free from assumptions. It employs a goal programming formulation to estimate the pre-harvest yield of sugarcane on the basis of measurements on biometrical characters of the plant. In order to assess the quality of forecasts, variance of residuals obtained from the proposed method has been compared with that obtained from the conventional regression analysis. The study reveals that there is no significant difference (P-value=0.43461) in the variances of the two residual series. Thus, without compromising the quality of forecast, the proposed alternative methodology can be adopted to estimate the sugarcane yield 3 months before harvest in situations, where the assumptions of conventional regression analysis are violated.