DATA CLASSIFICATION LEARNING METHOD, DATA CLASSIFICATION METHOD, DATA CLASSIFICATION LEARNER, DATA CLASSIFIER, STORAGE MEDIUM WITH DATA CLASSIFICATION LEARNING PROGRAM RECORDED, AND RECORDING MEDIUM WITH DATA CLASSIFICATION PROGRAM RECORDED

摘要

PROBLEM TO BE SOLVED: To improve the accuracy of data classification of test data even if the number of training data is small. SOLUTION: If the number of data with a classification class is zero, or if the ratio of positive examples in training data excels the ratio of positive examples in test data with a classification class, a positive example is input as the classification class of an test data item with a maximum evaluation value by a final classifier out of test data with no classification class. If not, a negative example is input as the classification class of a test data item with minimum evaluation by the final classifier out of the test data with no classification class.