PROBLEM TO BE SOLVED: To provide a method for using a support vector machine and a subset which form the subset of threshold data for discrimination by selecting a threshold suitable for the discrimination from a large amount of multivariable data stored in a memory by scanning that uses many sensors and many wavelengths to discriminate, classify and authenticate a document including paper money.SOLUTION: The method acquires a classification machine which classifies an item by collecting document data to execute discrimination and classification after setting a target value as a result of the subset, discrimination and classification of a plurality of threshold variables in accordance with a feature of the item, and evaluating compatibility reference while repeatingly changing a variable value about all remaining set variables until the number or compatibility of selected variables reaches a predetermined value.
展开▼