3.2. Statistical analysis and model fitting using RSMThe values of responses (yield of pectin) at different experimentalcombinations were given in Table 1. By employing multipleregression analysis on the experimental data, the predictedresponse Y for the yield of pectin could be obtained by the followingregression modelwhere Y is the yield of pectin (%) and X1, X2, X3 are the coded valuesof the test variables, power intensity of ultrasound (W/cm2); extractiontemperature (C); and sonication time (min), respectively.The statistical significance of regression equation was checkedby F-test, and the analysis of variance (ANOVA) for response surfacequadratic polynomial model was shown in Table 2. The value of the determination coefficient (R2) calculated from the quadraticregression model was 0.9855, which implied that 98.55% of thevariations could be explained by the fitted model. The value ofthe adjusted determination coefficient (R2Adj = 0.9668) is close toR2, indicating a high degree of correlation between the observedand predicted values. Besides, a low value of the coefficient of variation(C.V. = 0.95%) indicated a higher degree of precision and reliabilityof the experimental values.The corresponding variables would be more significant if theF-value becomes greater and the p-value becomes smaller, andvalues of p-value less than 0.05 showed model variables weresignificant (Atkinson & Donev, 1992). Therefore, the F-value(52.85) and p-value (<0.01) implied that this model was very significant.The lack-of-fit test measures the failure of the model torepresent the data at points that are not included in the regressionwithin the tested range. As shown in Table 2, F-value (4.68) and pvalue(0.0850) of lack-of-fit implied that it was not significantlyrelative to the pure error. All these results indicated that the modelwas adequate for predicting the yield of pectin under any combinationof the variables within the employed range. The significance ofeach coefficient was also determined using F-value and p-value:the linear coefficient (X1, X2), cross product coefficients (X1X3,X2X3) and quadratic term coefficients (X1 2 , X2 2 and X3 2 ) significantlyaffected the yield of pectin (p < 0.05).
การแปล กรุณารอสักครู่..