To provide a classifying apparatus capable of obtaining classification results in a way corresponding to loss values, even without preparing classifiers for all the subsets of attribute sets of measurements.
A data selecting part 2 creates numerical flat data without loss values and provides the data to a classifier creating part 3 along with classification categories, and determines whether or not the performance of each of the classifiers created based on the numerical flat data and the classification categories excels predetermined classification performance. A classifier storage part 4 stores the classifiers whose performance excels the predetermined classification performance, among a set of classifiers created by the classifier creating part 3. Based on the loss values of the group of measurements, a classifying part 5 obtains the classifiers selected from among the classifiers stored in the classifier storage part 4, and determines the classification category of the group of measurements provided.
Konobu Kato
Hideaki Tazawa
Hamada Hatsune