The LASSO Estimation Method for Linear EIV Model Parameters

DOI

10.2991/msbda-19.2019.4How to use a DOI?

Keywords

EIV model, Parameter estimation, Structural risk minimization principle, LASSO

Abstract

With regard to the linear EIV model, for the problem the weighted total least squares (WTLS) method only considers the goodness of fit, but ignores the complexity, which reducing its generalization ability, the LASSO estimation method for linear EIV model parameters (LE) was proposed that adding an L1 norm penalty to random error matrix of observation vector and coefficient matrix.Through the empirical study of the factors affecting the percentage of China's personal health expenditure in 2001-2017, compared with WTLS and least squares (LS) methods, the LE method could significantly improve the prediction accuracy, achieve stronger generalization ability and realize variable selection to achieve dimensionality reduction.

Copyright

© 2019, the Authors. Published by Atlantis Press.

Open Access

This is an open access article distributed under the CC BY-NC license (http://creativecommons.org/licenses/by-nc/4.0/).

Download article (PDF)

TY  - CONF
AU  - Mingqing Zhao
AU  - Tiantian Xi
PY  - 2019/08
DA  - 2019/08
TI  - The LASSO Estimation Method for Linear EIV Model Parameters
BT  - Proceedings of the 2019 International Conference on Modeling, Simulation and Big Data Analysis (MSBDA 2019)
PB  - Atlantis Press
SP  - 21
EP  - 27
SN  - 2352-538X
UR  - https://doi.org/10.2991/msbda-19.2019.4
DO  - 10.2991/msbda-19.2019.4
ID  - Zhao2019/08
ER  -