Scholarlink

<p>Dealing with excess zeros often poses challenges in model fitting when using traditional Poisson or other count regression models. To address these issues, hurdle (HP) and zero-inflated Poisson (ZIP) models offer alternative approaches. Specifically, the HP model separates the data into two distinct components: the hurdle and the non-zero count components, enabling separate modeling and estimation. This structure is particularly well-suited for the integration of shrinkage techniques, such as LASSO, to facilitate effective variable selection. The proposed hurdle LASSO (HP LASSO) not only produces a selection path and yields accurate results, but also demonstrates applicability to ZIP data under certain conditions in simulation studies. Moreover, the HP LASSO demonstrates practical efficacy in real-life scenarios, as evidenced by applications to the horseshoe crabs and traffic incidents datasets.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

On the use of the hurdle LASSO for modeling excess zero count data

  • 期刊论文

  • 发表日期：

    2026年6月3日

查看全文
  • Yang-Li Liao,
  • Chun-Yen Chen

摘要

Dealing with excess zeros often poses challenges in model fitting when using traditional Poisson or other count regression models. To address these issues, hurdle (HP) and zero-inflated Poisson (ZIP) models offer alternative approaches. Specifically, the HP model separates the data into two distinct components: the hurdle and the non-zero count components, enabling separate modeling and estimation. This structure is particularly well-suited for the integration of shrinkage techniques, such as LASSO, to facilitate effective variable selection. The proposed hurdle LASSO (HP LASSO) not only produces a selection path and yields accurate results, but also demonstrates applicability to ZIP data under certain conditions in simulation studies. Moreover, the HP LASSO demonstrates practical efficacy in real-life scenarios, as evidenced by applications to the horseshoe crabs and traffic incidents datasets.