تحليل الاتجاه و مشكلة تعدد العلاقة الخطية في تصميم القطع المجزأة

Abstract EN

Trend analysis was used for the yield of com crop of two factors by using of split plot design, the factor A (plant density) at three levels in the main plot which randomized complete block design applied, the factor B (fertilization) of three levels put in subplot.

The estimation results by using frend analysis method and ordinary least square methods are corresponding quite in establishment of the regression equation, in provision of appearance all experittient degrees were significant.

While dependent on trend analysis method become invalid if the regression equation contain unsignificant experiment degree.

The problem of inulticollinearity appeared, this problem is treated by using two methods: the first one was ridge regression which specify the degree of the factor and their interaction in effeeting the yield of corn, the second one was the principal components which went wrong under this experiment condition.