Fitting regression model with some assumptions for the identification of cotton yield factors

In agriculture and related fields many relationships exist that need to be identified in quantitative way. regression modeling plays an important role for the determination of such relationships and also the isolation of factors that greatly affect the target or response variable. for reliable and valid results,one has to check the regression assumptions like influential observations,multicollinearity etc. in this study,we have fitted the regression models with and without satisfying the some regression assumptions for the identification of cotton yield factors. for the analysis purposes,the required data was collected from the district khanewal. it was observed when regression assumptions were satisfied,model goodness (r2) was improved from 68% to 92%,r2 (adjusted) was improved from 62% to 90%) and standard error of the estimates reduced from 8.298 to 2.348. these better results indicated that the pesticide used for seed,all type of fertilizers (dap,potash and ganwara),water frequency,and previously sown crops were the significant factors for cotton yield with p-values as less than 0.05. so cotton yield was 90% explained by these factors.