Causal inference model construction method

A causal inference model construction method using a computer, wherein

• the computer comprises a control unit and a storage unit, and
• the control unit implements following steps of:
• a result value reception step of reading, from the storage unit, manufacturing log data including sensor data acquired in either or both of a manufacturing process and an inspection process;
• a correlation model construction step of constructing a correlation model using the manufacturing log data;
• a physical model reception step of receiving an input of a physical model obtained by modeling a physical law related to the manufacturing process or the inspection process;
• a probability distribution computation step of calculating probability distributions of the correlation model constructed in the correlation model construction step and the physical model;
• a probability distribution convergence processing step of performing convergence computation on the probability distribution of the correlation model, approximating a result of the convergence computation to the probability distribution of the physical model, and overwriting the correlation model;
• and
• a causal inference model holding step of storing a correlation model which is a processing result of the probability distribution convergence processing step in the storage unit as a causal inference model.