PURPOSE: To obtain the device which can automatically determine a timing classification result from time-series data, automatically introduce a feature quantity regarding a change with the lapse of time in each feature quantity, and explain the reason for the timing classification and prediction result to an operator, etc.
CONSTITUTION: This device is equipped with a classification class determining means 2 which determines a classification class on the basis of the change with the lapse of time of time-series data of a specific feature quantity showing the feature of one object system as to time-series data gathered by a time-series data gathering means 1, a symbolic feature quantity introducing means 3 which introduces a new symbolic feature quantity defined by the change with the lapse of time of each feature quantity of the time-series data from the means 1, a numerical feature quantity introducing means 4 which introduces a new numerical feature quantity defined by the change with the lapse of time of each feature quantity of the time-series data from the means 1, and a classifying and predictive calculating means 6 which inputs the outputs of the means 2, 3, and 4 respectively and calculates a classification predicted value of following data from data at a certain point of time.