階層化された評価項目のウエイト比に幅を与えた加重和評価法

摘要

複数の評価項目を有する対象の選択問題に対しては，評価項目にウエイト付けを行い加重和による評価を行うことが多い．ウエイト決定のためには評価対象を明確に把握することが困難であり，さらにこのような点を解決してウェイトを定めたとしてもウエイトの微小な変化が評価結果に多大な影響を及ぼす場合があることが指摘されており，得られた評価結果の妥当性を判断することが難しい．このため評価結果に対して感度分析を行うことが推奨されているが，評価項目の数が多く階層構造が構成されている場合については感度分析を行うことが困難となる．そこで本研究では感度分析を行うに際して，得られた評価項目のウェイト比に幅を与え各代替案の総合評価値を最大化することでウエイトを変化させ，評価結果に対する知見を得る方法を提案し，評価を行う際の一助となすことを目的とする．%This paper proposes a weight sum evaluation method that uses a range of weight ratios under a hierarchical structure of criteria and includes an added Data Envelopment Analysis (DEA) viewpoint. In a multi-criteria selection problem, one of the most popular methods to deal with this selection problem is called weighted sum evaluation. This evaluation means that a decision maker decides the criteria's weight and then calculates the weighted sum. The famous Analytic Hierarchy Process (AHP) method is used to solve this problem, which provides many theorems and cases. However, it is difficult to judge to what extent the result of this weighted sum is valid due to several reasons. One reason is that a decision maker must understand the problem to understand the correct weight. But in many cases, because time and cost are limited, such correct understanding is difficult. In addition, even if the decision maker can decide the weight under the circumstances, the alternative which was chosen by weight sum is not perfect because a little difference of the weight affects the evaluation result. Therefore, a sensitivity analysis is recommended in many cases. However, conducting this sensitivity analysis is still difficult because of the existence of many criteria and their hierarchical structure. This paper proposes a new method and applies an example of plant evaluation using the criteria of the Malcom Baldridge National Quality Award (MBNQA), and shows the advantage of proposing such a method.