L1正则化方法及其在经济增长中的应用

L1 Regularization Method and Its Application in Economic Growth Research

首先基于对L1正则化方法的介绍,将L1正则化应用到经济高质量增长模型中,用Lasso回归中的最小角回归方法筛选出对经济高质量增长影响较大的因素,并把影响较小的因素予以剔除。然后应用logistic回归和坐标下降法,对拥有较多维度的自变量进行筛选,以得到影响结果的主要维度,并对其他维度进行剔除。最后发现,劳动力受教育水平、产业结构对经济高质量增长的影响较大。

Based on the review of L1 regularization method, L1 regularization was applied to a high-quality economic growth model. The least-angle regression method in Lasso regression was used to screen out the factors, who had a greater impact on highquality economic growth, and eliminate the factors with less influence. Then we used logistic regression and coordinate descent method to screen independent variables with more dimensions, so as to obtain the main dimensions affecting the results and eliminate other dimensions. We finally found that, the education level of labor force and industrial structure both had greater impact on high-quality economic growth.