<![CDATA[The compilation of growth stand model usually uses the regression analysis. Homoscedasticity or residual kind homogeneity is one assumption which underlying the use of this regression analysis. Breaking this assumption causes the low of model accuracy which is shown by the low of determination coefficient and the height of error standard. The problem of heteroscedasticity can be solved by using weighted regression analysis.The Selected Raiser Growth Model equation in this research was transformed into a model equation: ln P = a + b/A, where there was a significant correlation between the growth and the age (R2 = 55.04%, sb0 = 0.041, and sb1 = 0.171). From the use of weighted regression analysis with weightier wi = 1/”Xi, it can be concluded that there was no real correlation between the growth and the age (R2 = 0.55%, sb0 = 0.572, and sb1 = 2.560). The use of weightier shows much lower accuracy than without weightier. However, from the use of weighted regression analysis with weightier: wi = 1/si2, where si2 = residual kinds at free variable group to I (X1) shows that there was significant correlation between the growth and the age (R2 = 45.46%; sb0 = 0.084, and sb1 = 0.205). There fore it can be said that the accuracy was much better than regression without weightier. Furthermore, the use of weighted regression analysis with weightier wi = 1/si2, where si2 is residual kind at free variable to i (X) which is estimated through second orde polynomial regression model shows a very significant correlation between the growth and the age (where R2 = 87.22%, sb0 = 0.029, and sb1 = 0.072). The last result shows a better accuracy than the preceding treatments. From this research, it can be concluded that by using a suitable weightier, the use of weighted regression analysis in compiling raiser growth model can improve the model accuracy.]]>
Copyrights © 2009
