Modelling Of Wheat Production In Punjab_through The Regularized Regression Approach While Addressing Multicollinearity

Agriculture sector plays a significant role in the economy of Pakistan. Amongst the major crops, wheat is essential food of Pakistani people's. Wheat production is a function of a number of factors. The identificatsion of important variables that affects the production of wheat is of prime interest of researcher to meet the basic need of increasing population. With multiple predictors, statistical modeling becomes complicated mainly due to the collinearity within the predictors. Regularized regression approaches do variable selection and shrinkage at same time. In this study, a rigorous comparison of the predictive performance of seven regularized regression approaches was performed via simulation while considering different levels of multicollinearity and sparsity. The results showed that Smoothly Clipped Absolute Deviation (SCAD) and Minimax Concave Penalty (MCP) performed better under low and high variation when the true model is sparse for all sample sizes based on mean squared error prediction (MSEP). Moreover, wheat production forecast model for Punjab province was estimated; while, using regularized regression methods and the significant predictors were identified. Among six predictors, area under crop, average retail price of fertilizer and average maximum temperature are important parameters that affect production of wheat. It is recommended that the farmers of the Punjab must be care full about these factors while sowing the wheat and government should provide facilities to farmers regarding these factors.