摘要 |
<p><P>PROBLEM TO BE SOLVED: To provide an energy demand prediction device, which is not affected by multicollinearity even though data causing the multicollinearity is used, and can represent the evidence of prediction results in an easy-to-understand manner for an operator to facilitate explanation of a prediction reason by adopting a prediction model similar to a multiple regression expression, while enhancing the prediction accuracy of energy demand. <P>SOLUTION: The energy demand prediction device constructs a prediction model with multicollinearity eliminated by a PLS (Partial Least Square) method, and uses such a prediction model to predict energy demand. <P>COPYRIGHT: (C)2010,JPO&INPIT</p> |