Method and program for model building equipment, as well as model building

摘要

PPROBLEM TO BE SOLVED: To construct a mixed model efficiently even when there are many model parameters. PSOLUTION: Geographic spatial data which have a plurality of cases that include a variable numerically representing the property of an evaluation object and position data representing a position in a geographic space, parameter information representing a parameter of a plurality of probability models obtained by modeling probability distribution of parameters, and application model information representing a probability model to be applied to each position in the geographic space are prepared. On the basis of the application model information, model dependency information numerically representing dependency between two probability models with respect to each combination comprising the same or two different probability models is calculated; a probability model to be applied to each position in the geographic space is selected, the application model information is updated, the parameters of the probability models corresponding to the respective plurality of groups to which the identical probability model is applied in the geographic spatial data is optimized, and the parameter information is updated so that the likelihood of the geographic spatial data may be increased with respect to a combination of parameters of the plurality of probability models and the model dependency information. PCOPYRIGHT: (C)2009,JPO&INPIT