| 摘要 |
<P>PROBLEM TO BE SOLVED: To improve prediction accuracy of data analysis. <P>SOLUTION: A data group is composed of a positive example data group of Ma pieces of positive example data and a negative example data group of Mb pieces of negative example data. (1) K pieces of positive example data are randomly sampled from the positive example data group. K pieces of negative example data are randomly sampled from the negative example data group. (2) Evaluation processing is performed by using the K pieces of positive example data and the K pieces of negative example data. (3) A weighted partially correct answer rate is obtained by executing the evaluation processing. The steps (1) to (3) are performed N times. N weighted partially correct answer rates are calculated by performing the steps (1) to (3) N times. (4) A weighted overall correct answer rate is calculated by averaging the N weighted partially correct answer rates. (5) As the weighted overall correct answer rate is calculated for each predictive algorithm, the weighted overall correct answer rate is graphed for each predictive algorithm. <P>COPYRIGHT: (C)2013,JPO&INPIT |
