勘察科学技术
勘察科學技術
감찰과학기술
SITE INVESTIGATION SCIENCE AND TECHNOLOGY
2015年
2期
46-48
,共3页
补偿最小二乘%时间序列%AR(p)模型%半参数%模型误差
補償最小二乘%時間序列%AR(p)模型%半參數%模型誤差
보상최소이승%시간서렬%AR(p)모형%반삼수%모형오차
penalized least squares method%time sequence%AR(p) model%semi-parametric%model error
处理时间序列常用的模型AR（ p）模型，它的核心是求取模型的参数。传统的参数求取方法忽略了模型误差的影响，为了顾及模型误差的影响，该文引入了半参数补偿最小二乘方法，将模型误差用非参数来加以弥补，提高了模型的精度。对比最小二乘方法、总体最小二乘方法和补偿最小二乘方法对时间序列的分析和预报结果。实验表明，补偿最小二乘方法所建立的模型对时间序列的分析和预报效果最好。
處理時間序列常用的模型AR（ p）模型，它的覈心是求取模型的參數。傳統的參數求取方法忽略瞭模型誤差的影響，為瞭顧及模型誤差的影響，該文引入瞭半參數補償最小二乘方法，將模型誤差用非參數來加以瀰補，提高瞭模型的精度。對比最小二乘方法、總體最小二乘方法和補償最小二乘方法對時間序列的分析和預報結果。實驗錶明，補償最小二乘方法所建立的模型對時間序列的分析和預報效果最好。
처리시간서렬상용적모형AR（ p）모형，타적핵심시구취모형적삼수。전통적삼수구취방법홀략료모형오차적영향，위료고급모형오차적영향，해문인입료반삼수보상최소이승방법，장모형오차용비삼수래가이미보，제고료모형적정도。대비최소이승방법、총체최소이승방법화보상최소이승방법대시간서렬적분석화예보결과。실험표명，보상최소이승방법소건립적모형대시간서렬적분석화예보효과최호。
The Ap( R) model is commonly used to process the time series ,and its core is calculating the model parameter .The traditional parameter calculating methods ignore the effect of model error .To con-sider the effect of model error ,in this paper ,the semi-parametric penalized least squares method is intro-duced.The model error is made up by non-parameter,the model accuracy is improved.By comparison with the analysis and prediction results on time series by the least squares method ,total least square meth-od and penalized least squares method , the experiment shows that the model built by penalized least squares get the best analysis and prediction results to time series .
