方法对比
并排查看您选择的方法；存在差异的行会高亮显示。
| 贝叶斯LASSO回归× | 弹性网络回归× | |
|---|---|---|
| 领域 | 统计学 | 统计学 |
| 方法族 | Regression model | Regression model |
| 起源年份≠ | 2008 | 2005 |
| 提出者≠ | Park & Casella | Hui Zou and Trevor Hastie |
| 类型≠ | Bayesian regularized regression | Penalized linear regression |
| 开创性文献≠ | Park, T., & Casella, G. (2008). The Bayesian Lasso. Journal of the American Statistical Association, 103(482), 681–686. DOI ↗ | Zou, H., & Hastie, T. (2005). Regularization and variable selection via the elastic net. Journal of the Royal Statistical Society: Series B (Statistical Methodology), 67(2), 301-320. DOI ↗ |
| 别名 | Bayesian LASSO, Bayesian L1 regression, double-exponential prior regression, Laplace prior regression | elastic net, EN regression, L1+L2 regularized regression, combined lasso-ridge regression |
| 相关≠ | 5 | 6 |
| 摘要≠ | Bayesian LASSO regression places double-exponential (Laplace) priors on regression coefficients, which is the Bayesian analogue of the classical LASSO penalty. It simultaneously shrinks small coefficients toward zero and performs soft variable selection, all within a coherent posterior inference framework that naturally quantifies parameter uncertainty through credible intervals. | Elastic net regression combines the L1 (lasso) and L2 (ridge) penalties into a single regularized regression framework. Controlled by a mixing parameter alpha and a shrinkage strength lambda, it can simultaneously select variables and handle correlated predictors — overcoming key limitations of pure lasso and pure ridge applied alone. |
| ScholarGate数据集 ↗ |
|
|