| 摘要 |
<P>PROBLEM TO BE SOLVED: To provide a test count determination method with which a suitable test count of tests can be determined when acquiring a characteristic value from test result of a test in which binary test result of success/failure can be obtained. <P>SOLUTION: Number (a) of characteristic values to be acquired for each set, number (b) of sets and temporary normal distribution representing scatter of characteristic values for each test are set in advance. Random numbers according to the temporary normal distribution are generated as many as the number (a) and a mean value (Ms) thereof and a standard deviation (Ss) thereof are determined iteratively as many times as the number (b) of sets. A mean value and a standard deviation of the mean value (Ms) and a mean value and a standard deviation of the standard deviation (Ss) are determined. On the other hand, random numbers according to the temporary normal distribution are generated as many as number (c), test result in performing tests (c) times is predicted and a characteristic value based on the result is determined iteratively (a) times to determine a mean value (Mt) and a standard deviation (St) iteratively as many times as the number (b) of sets. A mean value and a standard deviation of the mean value (Mt) and a mean value and a standard deviation of the standard deviation (St) are determined, the mean values and the standard deviations of both are tested by a statistical test and the statistical test is repeated while changing a test count (c), thereby determining the test count (c) in which absence of a significant difference is determined. <P>COPYRIGHT: (C)2013,JPO&INPIT |
