System and method for calculating generalized utilities and choice predictions

A non-transitory computer-readable medium having program code recorded thereon for calculating generalized utilities and choice predictions, the program code being executed by a processor and comprising:

• program code to identify an individual's choice a user desires to predict and relevant parameters influencing the individual's choice;
• program code to analyze, using a natural language processor (NLP), language communications from a data log to determine the effects of social distance and personal distance on the individual's choice;
• program code to analyze, using a machine learning (ML) model, non-language communications from a sequence of interaction logs to determine compromises, risks, and costs on the individual's choice;
• program code to provide a machine learning (ML)-based estimation of different function forms and relevant parameters for an expected generalized utility (EGU) model according to an outcome parameter and an uncertainty parameter based on the analyzing of the non-language communications, and a social distance parameter and a temporal distance parameter based on the analyzing of the language communications regarding the individual's choice;
• program code to integrate the estimates of the function forms and the relevant parameter estimates to configure the EGU model;
• program code to predict, by the EGU model, the individual's choice as well as a confidence interval associated with the predicted, individual's choice;
• and
• program code to display, through a user interface, the predicted, individual's choice and the associated confidence interval.

A system for calculating generalized utilities and choice predictions, the system comprising:

• a choice identification module to identify an individual's choice a user desires to predict and relevant parameters influencing the individual's choice;
• a natural language processor (NLP) language to analyze communications from a data log to determine the effects of social distance and personal distance on the individual's choice;
• an estimated EGU parameter/function module to analyze, using a machine learning (ML) model, non-language communications from a sequence of interaction logs to determine compromises, risks, and costs on the individual's choice and to provide a machine learning (ML)-based estimation of different function forms and relevant parameters for an expected generalized utility (EGU) model according to an outcome parameter and an uncertainty parameter based on the analyzing of the non-language communications, and a social distance parameter and a temporal distance parameter based on the analyzing of the language communications regarding the individual's choice;
• an EGU parameter/function integration model to integrate the estimates of the function forms and the relevant parameter estimates to configure the EGU model;
• the EGU model to predict the individual's choice as well as a confidence interval associated with the predicted, individual's choice;
• and
• a user interface to display the predicted, individual's choice and the associated confidence interval.