摘要 |
PURPOSE: A cluster result analysis system of massive data and a method thereof are provided to calculate a silhouette coefficient, thereby objectively analyzing the clustering result of massive data by applying Hadoop. CONSTITUTION: A task management device (200) divides the clustered object file into the designated size of blocks, combines the blocks, and generates an input split. A distance calculation device (210) receives an allocation of the input split, and calculates the distance sum of each record from block included in the input split. An index coefficient calculation device (220) uses the distance sum of each record and calculates a silhouette coefficient of each record. An analysis device (230) calculates the average of the silhouette coefficient of each record and calculates the final silhouette coefficient of the cluster. [Reference numerals] (200) Task management device; (210,AA,CC) Distance calculation device; (220,BB,DD) Index coefficient calculation device; (230) Analysis device |