2001 年 4 巻 p. 33-40
In recent years, data accumulationh as been remarkablyi ncreased, and it is pointed out that it is necessaryt o restored ata in order, and to discoverr ulesh iddenb ehindd ata. The notion of data mining is firstc larifieda nd is interpreteda s a non-structurailn verseo ridentificationp roblem. Then, within the scope of data mining, n umerical analysesare carried out to find out rules on three sets of data whose featuresare different, that is, (1) occurrence of ared tide in Tokyo bay, (2) causalr elationship of water qualityin Lake Shinji and Nakaumi. Finally, aneural network algorithm is found that the usefulness of a data mining procedure for non-structural system identification.