遺伝的アルゴリズムを利用した項目選択

概要

論文の詳細を見る
The items in a test or questionnaire must often be selected to fulfill, as much as possible, some contradictory requirements in the psychological measurement. For example, sometimes the number of items in the scale must be as small as possible in order to maintain the reliability of scale when measuring a characteristic. Under such a situation, item selection generally depends on trial and error in which unavailable items are deleted one by one from the items in the scale while considering the level so that the reliability does not decrease too much. This process requires time and effort; moreover, in this process the intentions of the researcher selecting the items may greatly affect the result. In this study, the items were automatically selected using a genetic algorithm. As a result, the validity of the model in which the genetic elite and the exactly opposite non-elite were saved at the same time was confirmed.
文教大学の論文
1999-12-20