METHOD FOR OPTIMIZING PREDICTIVE ALGORITHM BASED EMPIRICAL MODEL

摘要

Disclosed is a method for optimizing a predictive algorithm based on an empirical model. The method for optimizing a predictive algorithm based on an empirical model comprises the steps of: allowing a control unit to group training data stored in a training data DB based on a correlation between sensor information and the training data; allowing the control unit to generate the plurality of training data by dividing the grouped training data by a plurality of set division lengths; allowing the control unit to generate each prediction data from the divided training data and compare errors with test data; and allowing the control unit to generate optimal prediction data by deriving a weight and a set division length at which an error value is minimum as a result of comparing the errors.