Author(s): B. GURUNG, S. DUTTA, K.N. SINGH, A. LAMA, S. VENNILA AND B. GURUNG

Abstract: Regression analysis is one of the most commonly employed statistical tool for analyzing the “cause and effect” relationship. The regression coefficients are tested to find if they are affecting the dependent variable and the sign and the values of the coefficients help us to know how and by much they are affecting the dependent variables. However, this technique may not be appropriate for statistical analysis where the dependent variable is in proportional scale as they are restricted to the interval (0, 1). As such, the assumptions of error terms being normal and homoscedastic are violated.A reasonable alternative is to use transformations, but this again may lead to bias in estimation. Taking this into consideration, researchers have developed techniques to capture the changing proportional data such that interpretation is easier and also it becomes more flexible than transformation. In our research, we developed a statistical model for forewarning Helicoverpa armigera infestation by employing beta regression. By maximizing the likelihood function by employing the “optim” function through the analytical gradients available in R package, the unknown coefficients are estimated.Moreover, Fisher scoring iteration through the use of expected information and analytical gradients were used to obtain better estimates through the use of “optim”. Residual diagnostics have also been carried out. The developed model could also be employed for forewarning pest infestation in other crops.

DOI: 10.22271/09746315.2023.v19.i1.1681

Download Full Article: Click Here