1997 年 5 巻 6 号 p. 45-52
In the previous paper, the approach of periodical health exanimation using Mahalanobis Distance without missing data was introduced, When there is missing data, the Mahalanobis Space cannot be accurately constructed and the Mahalanobis Distance calculated may lead a wrong conclusion. In this paper, data on sex, age and 25 blood biological tests were used to construct a Mahalanobis Space, Missing data were systematically introduced as the real data were replaced by zeros. The differences in diagnosis results were observed. The results showed that even a small fracton of missing data will significantly change the judgment. ln the next paper, some countermeasures for missing data will be proposed and the results reported.