Linear and Non-Linear Regression Techniques to Develop Predictive Models for Pearl Millet and Finger Millet Trends of Productivity in India

Abstract

In this study, we employed linear and non-linear regression models to determine the optimal trend for pearl millet and finger millet, considering their respective areas, production, and productivity. For this study, we have summarized the secondary annual data according to area, production, and productivity. We have used various regression models, including linear, quadratic, cubic, and logarithmic models, to determine the best trend. After analyzing these models and validating the model, we get a cubic model that shows the best trend for pearl millet area, production, and productivity with the value of coefficient of determination was 86 percentages, 60 percentages, 90 percentages for the area, production, and productivity, respectively, the value of the coefficient of determination shows how the model performed best with higher accuracy. In the finger millet crop, to get the best trend model, we found that the quadratic model and cubic model showed almost similar results, but some validation parameters showed that the cubic model was fitted. Overall, we can say the cubic model was the best for both crops.