The model defines four attributes: WWT, PWT, RANK and ERROR. WWT and PWT are used to store the utilities of the warranty and price configurations, respectively. The model will solve for values of WWT and PWT that minimize total prediction error. These values will then give us the implied utility functions for the two product attributes. RANK is used to store the customer preference rankings. Finally, ERROR stores the error in predicting the preference of one configuration over another given the implied utility values contained in WWT and PWT.