To easily predict data or an arbitrary index value by calculating the characteristic value of a data population from input data based on a prescribed relational formula to analytically estimate the characteristic value of the data population without being classified by cases.

A data conversion part 13 converts input data transferred from an input part 11 to secondary data. A characteristic value estimating part 15 uses the least square method to estimate the characteristic value of the data population from input data based on a prescribed equation derived from fundamental models including a Gompertz curve model and a logistic curve model. A prediction formula determining part 19 determines a prediction formula in accordance with the characteristic value, whose validity is verified by a verification part 17, and input data. A data prediction part 21 predicts data of an arbitrary index value of the data population based on the determined prediction formula.