<資料>入学試験教科の重みづけ計算法について

概要

論文の詳細を見る
We analyze the student grade in Yamanashi university (U-grade) and that of entrance examination (E-grade) and try to get optimal weight for each subject in entrance examination. The weight is assumed to be optimal when weighted E-grade has maximal correlation with U-grade. At first, we assume several nonweighted vectors as U-grade and try to get optimal weight by multiple correlation method. Because this weight is not practical, we assume weighted U-grade and E-grade, and calculate optimal weight by canonical correlation method. But, results are also unsuccessful. To investigate these results, we analyze statistical structure of both grades and find that it is not adequate to assume only one vector to represent each grade. Probably, this is the main reason why simple correlation method does not work well.
山梨大学の論文