当前位置: X-MOL 学术J. Korean Stat. Soc. › 论文详情
Our official English website, www.x-mol.net, welcomes your feedback! (Note: you will need to create a separate account there.)
Zero-inflated Poisson-Akash distribution for count data with excessive zeros
Journal of the Korean Statistical Society ( IF 0.6 ) Pub Date : 2023-06-16 , DOI: 10.1007/s42952-023-00216-5
Mohammad Kafeel Wani , Peer Bilal Ahmad

Over-dispersed models are often used whenever the variation is more than what in point of fact is anticipated by a model. One of the reasons behind experiencing over-dispersion is an excessive number of zeros, hence when modeling this observed fact, we use zero-inflated models rather than more traditional ones. As a part of our research, we have suggested a zero-inflated variant of Poisson-Akash distribution that was introduced in 2015. We have calculated crucial statistical characteristics of the suggested model which are not confined to generating functions, over-dispersion property, moments and associated measures. The parametric estimation has been carried out using the maximum likelihood estimation. Two different simulation exercises have been carried out, one to test the performance of maximum likelihood estimates and the other for testing the compatibility of our devised model when data has been simulated from different zero-inflated models. For the purpose of testing the compatibility of our proposed model, we have used four real life data sets and considered different performance measures like Goodness-of-fit, Akaike’s information criterion, Bayesian information criterion, Dispersion index etc. The fitting results have been compared with some other models of interest. Moreover, we have tested the significance of the zero-inflation parameter using Likelihood ratio test, Score test and the Wald test itself.



中文翻译:

具有过多零的计数数据的零膨胀泊松-阿卡什分布

当变化大于模型实际预期的变化时,通常会使用过度分散模型。经历过度离散的原因之一是零数量过多,因此在对这一观察到的事实进行建模时,我们使用零膨胀模型而不是更传统的模型。作为我们研究的一部分,我们提出了 2015 年引入的 Poisson-Akash 分布的零膨胀变体。我们计算了所建议模型的关键统计特征,这些特征不仅限于生成函数、过度离散特性、矩及相关措施。使用最大似然估计进行参数估计。进行了两种不同的模拟练习,一个用于测试最大似然估计的性能,另一个用于在从不同的零膨胀模型模拟数据时测试我们设计的模型的兼容性。为了测试我们提出的模型的兼容性,我们使用了四个现实生活数据集,并考虑了不同的性能指标,如拟合优度、赤池信息准则、贝叶斯信息准则、离散度指数等。对拟合结果进行了比较与一些其他感兴趣的模型。此外,我们还使用似然比检验、Score检验和Wald检验本身检验了零通胀参数的显着性。我们使用了四个现实生活数据集,并考虑了不同的性能指标,如拟合优度、Akaike 信息准则、贝叶斯信息准则、离散度指数等。拟合结果已与其他一些感兴趣的模型进行了比较。此外,我们还使用似然比检验、Score检验和Wald检验本身检验了零通胀参数的显着性。我们使用了四个现实生活数据集,并考虑了不同的性能指标,如拟合优度、Akaike 信息准则、贝叶斯信息准则、离散度指数等。拟合结果已与其他一些感兴趣的模型进行了比较。此外,我们还使用似然比检验、Score检验和Wald检验本身检验了零通胀参数的显着性。

更新日期:2023-06-21
down
wechat
bug