方法对比

	Lasso 回归	面板数据固定效应模型	泊松回归与负二项回归
领域	机器学习	计量经济学	计量经济学
方法族	Machine learning	Regression model	Regression model
起源年份	1996	2014	1998
提出者	Tibshirani, R.	Hsiao (textbook treatment); within transformation of panel data	Cameron & Trivedi (textbook treatment); Hilbe (negative binomial)
类型	Regularized linear regression (L1 penalty)	Panel data regression	Generalized linear model for count data
开创性文献	Tibshirani, R. (1996). Regression Shrinkage and Selection via the Lasso. Journal of the Royal Statistical Society: Series B, 58(1), 267–288.	Hsiao, C. (2014). Analysis of Panel Data (3rd ed.). Cambridge University Press.	Cameron, A. C. & Trivedi, P. K. (1998). Regression Analysis of Count Data. Cambridge University Press.
别名	LASSO Regresyonu, lasso, L1-regularized regression, L1 regularization	fixed effects model, within estimator, panel fixed-effects regression, Panel Veri — Sabit Etkiler Modeli	count regression, log-linear count model, negative binomial regression, Poisson / Negatif Binom Regresyon
相关	4	5	4
摘要	Lasso regression, introduced by Robert Tibshirani in 1996, is a linear regression method that adds an L1 penalty to the loss so that it shrinks coefficients and performs variable selection at the same time, producing a sparse model. By driving some coefficients exactly to zero it keeps only the predictors that matter.	The Panel Data Fixed Effects model estimates relationships from panel data (the same units observed over several time periods) while controlling for unit- and/or time-specific effects, supporting causal inference. It is developed as the within estimator in standard treatments such as Hsiao's Analysis of Panel Data (2014).	Poisson regression is a generalized linear model for count outcomes — events tallied as non-negative integers such as hospital admissions, accidents, or article counts. It models the log of the expected count as a linear function of the predictors, and is developed in the standard count-data treatment of Cameron and Trivedi (1998); when the counts are over-dispersed, the closely related negative binomial model (Hilbe, 2011) is preferred.