Model update device, method, and program

A model update method, comprising:

• acquiring, by a processor, an explanatory variable that is to be input to a model represented by a tree structure including a plurality of nodes, the model being configured to output an objective variable of a control target for the explanatory variable;
• associating, by the processor, a frequency at which an explanatory variable included in each of a plurality of areas, which are obtained by dividing an explanatory variable space, is acquired with each of the plurality of areas, and specifying an area to which an explanatory variable included in learning data used to learn the model belongs and in which a frequency of an explanatory variable is a predetermined value or less;
• updating, by the processor, the model by removing nodes in the model that have been trained with learning data including an explanatory variable belonging to the specified area;
• and
• generating a control signal from an objective variable that is output from the model and controlling the control target by use of the control signal.

A non-transitory storage medium storing a model update program that causes a computer including a processor to cause the processor to:

• acquire an explanatory variable that is to be input to a model represented by a tree structure including a plurality of nodes, the model being configured to output an objective variable of a control target for the explanatory variable;
• associate a frequency at which an explanatory variable included in each of a plurality of areas, obtained by dividing an explanatory variable space, is acquired with each of a plurality of areas, and specifying an area to which an explanatory variable included in learning data used to learn the model belongs and in which a frequency of an explanatory variable is a predetermined value or less;
• update the model by removing nodes in the model that have been trained with learning data including an explanatory variable belonging to the specified area;
• and
• generate a control signal from an objective variable that is output from the model and control the control target by use of the control signal.