| 摘要: |
| 目前,长期预报的业务工作主要还是采用统计方法,传统的统计建模要求建模之前建模者必须事先确定或假设模式里的应变量和变量之间的函数关系f,而统计建模工作的实质是根据数据或资料去确定该函数f里的参数或系数。 |
| 关键词: |
| DOI： |
| 分类号: |
| 基金项目: |
|
| THE SELF-ORGANIZING MODELING METHOD FOR LONG-TERM FORECAST——ROOT MEAN SQUARE AS CRITERION |
|
Lin Zhenshan
|
|
Geophysics Department of Beijing University
|
| Abstract: |
| The Group Method of Date Handling (GMDH) algorithm is applied to construct a nonlinear model for long-term forecast. Being a "self-organizing" active high-order nonlinear regressive modeling technique, GMDH is more objective and requires less data as conpared with traditional" passive" statistic modeling methods. It promises new hope for long-term statistic prediction, interpolation and extension of nonlinear time serials. |
| Key words: |
