To efficiently form a cluster (partial data) with a sufficient evaluation value.
This cluster analysis system is provided with a first means inputting partial data groups formed by sorting a plurality of data according to similarity, a second means displaying the contents of the inputted partial data into a visually recognizable form, a third means indicating specified partial data among the displayed partial data, a forth means forming new partial data including the indicated partial data, and a fifth means calculating an evaluation value expressing conformity or nonconformity of a data distribution of the data included in the new partial data and the data of the other partial data and displaying the evaluation value into the visually recognizable form.
YAMAMOTO NORIYUKI
YAMASHITA IWAO
HASHIDA JUNYA