Effect of different biochemical traits on seed cotton yield: An application of liu linear regression

Abstract

This article aimed to study the associations among the biochemical traits and their effects on seed cotton yield using the regression analysis and to assess the alternative approach for reducing the impact of multicollinearity problem in estimating the regression coefficients. The field experiment was conducted where five explanatory variables (chlorophyll ‘a’, chlorophyll ‘b’, total chlorophyll, total soluble protein and total soluble sugar) and one dependent variable seed cotton yield were measured. The correlation matrix of showed that biochemical traits were significantly correlated. The multicollinearity problem among the biochemical traits was determined by condition index and correlation matrix. Using the least square regression analysis, the effects of biochemical traits on seed cotton yield were not satisfactory since least square regression model has high value of MSE (3352475), AIC (366.7) and inconsistent estimates of traits. The Liu regression analysis was efficient (MSE = 57212 and AIC = 363.8) and reliable in reducing the adverse effects of multicollinearity. The Liu regression results indicated that total chlorophyll and total soluble protein were contributed a significant (P ≤ 0.05) role in seed cotton yield. In contrast, ordinary least square regression analysis was showed insignificant (P > 0.05) effect of total chlorophyll on seed cotton yield.