Abstract

Food security has been considered as a major concern for many countries since several years. Food is basic necessity of human beings all over the world. World’s Population and demand of food is increasing across the years while its production is not enough to meet this challenge of food security. Wheat is paramount food crop all over the world. Main objectives of this study is to identify data set (data units form) that could provide a model with better prediction capability and to investigate the significant factors for wheat yield enhancement. Hierarchical regression analysis is applied on the data taken from Crop Reporting Service, Agriculture Department of Punjab, Pakistan. Three more data sets (clusters) generated from the original data set. Model selection criteria, adjusted R2, ΔR2, MSE AIC, SIC, Wi (AIC) and ER (AIC) have been exercised on these models. The result indicates that clustering improved the R2, Cronbach's alpha and reduced the variance, MSE, AIC, SIC. The best model is selected on the basis of prediction capability and it can be helpful for precise estimation of food to cope with the coming challenge of food security.

Key words: Hierarchical regression, multiple regressions, weighted least squares, wheat productivity