A simulation study for count data models under varying degrees of outliers and zeros

Tüzen, Fatih; Erbaş, Semra; Olmuş, Hülya

doi:10.1080/03610918.2018.1498886

Cited by 12 publications

(5 citation statements)

References 25 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This is in line with the results of our simulation study stating that when the overdispersion occur, the ZIP model performed relatively the same or better than the Hurdle Poisson model under different scenarios, but ZIP and Hurdle models are better than the standard Poisson model for all given n based on AIC. These results are in agreement with some other studies stating that Zero Inflated and Hurdle models are similar in terms of AIC [5,6]. Another study also concluded a result somewhat similar to that found in our study that the ZIP model performed better than the Poisson hurdle (PH) model [10].…”

Section: B Case Studysupporting

confidence: 93%

“…A number of studies have compared the Zero Inflated and Hurdle models. For example, research has concluded that Zero Inflated and Hurdle models are indiscernible in terms of goodness of fit measures namely the Akaike information criterion (AIC) [5,6]. Research has compared different counts models including ZIP, Hurdle Poisson (HP), zero-inflated negative binomial (ZINB), hurdle negative binomial (HNB), Poisson, and negative binomial (NB) models for analyzing inpatient hospitalization data of patients' psychiatric disorders [7].…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Evaluating the Performance of Zero-Inflated and Hurdle Poisson Models for Modeling Overdispersion in Count Data

Aswi

Astuti²,

Sudarmin

2022

Inferensi

View full text Add to dashboard Cite

A Poisson regression model is commonly used to model count data. The Poisson model assumes equidispersion, that is, the mean is equal to the variance. This assumption is often violated. In count data, overdispersion (the variance is larger than the mean) occurs frequently due to excessive zeroes in the response variable. Zero-inflated Poisson (ZIP) and Hurdle models are commonly used to fit data with excessive zeros. Although some studies have compared the ZIP and Hurdle models, the results are inconsistent. This paper aims to evaluate the performance of ZIP and Hurdle Poisson models for overdispersion data through both simulation study and real data. Data were simulated with three different sample sizes, six different means, and three different probabilities of zero with 500 replications. Model goodness-of-fit measures were compared by using Akaike Information Criteria (AIC). Overall, the ZIP model performed relatively the same or better than the Hurdle Poisson model under different scenarios, but both ZIP and Hurdle models are better than the standard Poisson model for overdispersion in count data.

show abstract

Section: B Case Studysupporting

confidence: 93%

Section: Introductionmentioning

confidence: 99%

Evaluating the Performance of Zero-Inflated and Hurdle Poisson Models for Modeling Overdispersion in Count Data

Aswi

Astuti²,

Sudarmin

2022

Inferensi

View full text Add to dashboard Cite

show abstract

“…Similarly, Nekesa et al (2019) reported results from simulation study and real data analysis of exposed infant diagnosis, showing the negative binomial emerging as the best performing model when fitting data with both structured and non-structured zeros under various settings. Tüzen et al (2018) reported simulation scenarios, where the NB model outperformed other count models in the presence of outliers and/or excess zeros. Allison (2012) noted that some applications exist in which a compelling case could be made for a zero-inflated model and suggested the use of ZINB instead of ZIP when modeling zero-inflated count data.…”

Section: Discussionmentioning

confidence: 99%

“…Fernandez and Vatcheva BMC Medical Research Methodology (2022) 22:211 Tuzen et al (2018) examined the performance in terms of fit of Poisson, NB, ZIP, ZINB, Poisson Hurdle and NB Hurdle models under various outliers and zero-inflation scenarios of simulated data and found that ZINB and NB Hurdle were superior to Poisson, NB, and ZIP models. They also reported that in some scenarios, the NB model outperformed all models in the presence of outliers and/or excess zeros [44]. Tlhaloganyang et al [45] compared NB with ZIP and ZINB models using different real datasets characterized by overdispersion and zero-inflation.…”

Section: Introductionmentioning

confidence: 99%

A comparison of statistical methods for modeling count data with an application to hospital length of stay

Fernández

Vatcheva

2022

BMC Med Res Methodol

View full text Add to dashboard Cite

Background Hospital length of stay (LOS) is a key indicator of hospital care management efficiency, cost of care, and hospital planning. Hospital LOS is often used as a measure of a post-medical procedure outcome, as a guide to the benefit of a treatment of interest, or as an important risk factor for adverse events. Therefore, understanding hospital LOS variability is always an important healthcare focus. Hospital LOS data can be treated as count data, with discrete and non-negative values, typically right skewed, and often exhibiting excessive zeros. In this study, we compared the performance of the Poisson, negative binomial (NB), zero-inflated Poisson (ZIP), and zero-inflated negative binomial (ZINB) regression models using simulated and empirical data. Methods Data were generated under different simulation scenarios with varying sample sizes, proportions of zeros, and levels of overdispersion. Analysis of hospital LOS was conducted using empirical data from the Medical Information Mart for Intensive Care database. Results Results showed that Poisson and ZIP models performed poorly in overdispersed data. ZIP outperformed the rest of the regression models when the overdispersion is due to zero-inflation only. NB and ZINB regression models faced substantial convergence issues when incorrectly used to model equidispersed data. NB model provided the best fit in overdispersed data and outperformed the ZINB model in many simulation scenarios with combinations of zero-inflation and overdispersion, regardless of the sample size. In the empirical data analysis, we demonstrated that fitting incorrect models to overdispersed data leaded to incorrect regression coefficients estimates and overstated significance of some of the predictors. Conclusions Based on this study, we recommend to the researchers that they consider the ZIP models for count data with zero-inflation only and NB models for overdispersed data or data with combinations of zero-inflation and overdispersion. If the researcher believes there are two different data generating mechanisms producing zeros, then the ZINB regression model may provide greater flexibility when modeling the zero-inflation and overdispersion.

show abstract

“…Among these studies, the conclusions are inconsistent. Some revealed that ZI and hurdle models are indistinguishable with respect to goodness of fit measures ( Xu et al 2015 ; Tüzen et al 2018 ); whereas, some studies found the hurdle model had a better fit than the ZI model ( Min and Agresti 2005 ; Sharker et al 2020 ) and other empirical application found ZI model performs better than the hurdle model ( Hu et al 2011 ). It is therefore desired to identify the situations where hurdle models perform better than ZI and vice versa through simulation studies.…”

Section: Introductionmentioning

confidence: 99%

A comparison of zero-inflated and hurdle models for modeling zero-inflated count data

Feng

2021

J Stat Distrib App

148

View full text Add to dashboard Cite

Counts data with excessive zeros are frequently encountered in practice. For example, the number of health services visits often includes many zeros representing the patients with no utilization during a follow-up time. A common feature of this type of data is that the count measure tends to have excessive zero beyond a common count distribution can accommodate, such as Poisson or negative binomial. Zero-inflated or hurdle models are often used to fit such data. Despite the increasing popularity of ZI and hurdle models, there is still a lack of investigation of the fundamental differences between these two types of models. In this article, we reviewed the zero-inflated and hurdle models and highlighted their differences in terms of their data generating processes. We also conducted simulation studies to evaluate the performances of both types of models. The final choice of regression model should be made after a careful assessment of goodness of fit and should be tailored to a particular data in question.

show abstract

A simulation study for count data models under varying degrees of outliers and zeros

Cited by 12 publications

References 25 publications

Evaluating the Performance of Zero-Inflated and Hurdle Poisson Models for Modeling Overdispersion in Count Data

Evaluating the Performance of Zero-Inflated and Hurdle Poisson Models for Modeling Overdispersion in Count Data

A comparison of statistical methods for modeling count data with an application to hospital length of stay

A comparison of zero-inflated and hurdle models for modeling zero-inflated count data

Contact Info

Product

Resources

About