李坤, 王潜心, 闵扬海,等.附有周期项的二次多项式LASSO钟差预报模型[J].河南理工大学学报（自然科学版）,2022,41(3):74-80.

LI K, WANG Q X, MIN Y H, et al.A clock offset prediction model with quadratic polynomial based on LASSO algorithm[J].Journal of Henan Polytechnic University(Natural Science) ,2022,41(3):74-80.

附有周期项的二次多项式LASSO钟差预报模型

李坤1, 王潜心1, 闵扬海2, 龚佑兴3, 苗伟1, 程彤1

1.中国矿业大学 环境与测绘学院，江苏 徐州 221116;2.苏州市房地产市场与交易管理中心，江苏 苏州 215002;3.国防科学技术大学 指挥军官基础教育学院，湖南 长沙 410072

摘要:为了解决最小二乘估计（least squares estimation ， LSQ）算法在处理高维度数据模型式时易产生模型过拟合、预报精度不高等问题，采用LASSO算法（least absolute shrinkage and selection operator）对附有周期项的二次多项式模型进行整体求解，分析6，12，18，24 h预报精度。结果表明，LASSO算法能有效避免模型参数求解的过拟合问题，极大提高二次多项式模型的预报精度，随着预报时间增加，LASSO算法优势愈加明显。

关键词:二次多项式;周期项;LASSO;最小二乘估计;过拟合

doi:10.16186/j.cnki.1673-9787.2020090051

基金项目:国家自然科学基金资助项目（41874039）；江苏省自然科学基金资助项目（BK20191342）

收稿日期:2020/09/11

修回日期:2020/11/03

出版日期:2022/05/15

A clock offset prediction model with quadratic polynomial based on LASSO algorithm

LI Kun1, WANG Qianxin1, MIN Yanghai2, GONG Youxing3, MIAO Wei1, CHENG Tong1

1.School of Environment and Geoinformatics ， China University of Mining and Technology， Xuzhou  221116 ， Jiangsu， China;2.Suzhou Real Estate Market and Transaction Management Center， Suzhou  215002 ， Jiangsu， China;3.School of Military Commanding Officer Basic Education，National University of Defense Technology，Changsha  410072，Hunan，China

Abstract: In order to solve the problems of model over-fitting and low prediction accuracy of the least squares estimation (LSQ) in processing high-dimensional data model formulas,the LASSO algorithm was used to solve the overall model of the quadratic polynomial model with periodic terms , and 6 ,12 ,18,24 h accuracy analysis of the prediction experiment respectively was carried out. The result showed that the LASSO algorithm could effectively avoid the over-fitting phenomenon in the solution of model parameters , greatly improve the prediction accuracy of the quadratic polynomial, and as the prediction time increased,the advantages of the lasso algorithm increased obviously.

Key words:quadratic polynomial;period term;LASSO;least squares estimation;over-fitting

附有周期项的二次多项式LASSO钟差预报模型_李坤.pdf