Regression Models for Count Data in<i>R</i>

Zeileis, Achim; Kleiber, Christian; Jackman, Simon

doi:10.18637/jss.v027.i08

Cited by 1,948 publications

(1,485 citation statements)

References 19 publications

Supporting

Mentioning

1,413

Contrasting

Unclassified

Order By: Relevance

“…The data are said censored if a range of values of the dependent variable is collapsed where by responses might be 0, 1, 2, and 3 or more (Zeileis et al, 2008;Greene, 2002). Such situation is common in clinical work whereby one often encounters situations where the outcome variable is numeric, but in the form of counts.…”

Section: Resultsmentioning

confidence: 99%

Farmers' Preferential Choice Decisions to Alternative Cassava Value Chain Strands in Morogoro Rural District, Tanzania

Sewando¹,

Mdoe²,

Mutabazi³

2011

Agricultural J.

View full text Add to dashboard Cite

Abstract:The study was conducted to determine farmers' preferential choice decision of alternative cassava value chain strands and as well as factors behind such decisions in Morogoro rural District, Tanzania. Factor analysis was first used to reduce and identify the factors (variables for further analysis whereby the factors with highest eigen-value were applied to develop factor scores to measure the attitudinal variables. Results indicate that farmers have positive risk attitude towards participation in the alternative cassava value chain strands for commercialization. A count data model known as Poisson model was applied to determine the factors which influenced this attitude. Results indicated that farm size, experience, femaleheaded households and land-holding had influenced the farmers' preferential choice decision.Recommendations for enhancing farmers' participation in profitable cassava value chain strands are strengthening coordination, provision of improved cassava varieties and introduction of cassava processing technologies.

show abstract

Section: Resultsmentioning

confidence: 99%

Farmers' Preferential Choice Decisions to Alternative Cassava Value Chain Strands in Morogoro Rural District, Tanzania

Sewando¹,

Mdoe²,

Mutabazi³

2011

Agricultural J.

View full text Add to dashboard Cite

show abstract

“…In addition, zeroes occur more frequently in the data than predicted by a Poisson model (that is, most people were not in the hospital or emergency room for most observation months). We accommodated this overrepresentation by using a zero-inflated model (18, 19). This model entailed a two-part regression that both tested the probability of having no visits (a binomial model) and then predicted a count of the number of visits.…”

Section: Methodsmentioning

confidence: 99%

Impact of Illness Management and Recovery Programs on Hospital and Emergency Room Use by Medicaid Enrollees

Salyers

Rollins²,

Clendenning³

et al. 2011

Psychiatric Services

View full text Add to dashboard Cite

Objective Illness management and recovery is a structured program that helps consumers with severe mental illness learn effective ways to manage illness and pursue recovery goals. This study examined the impact of the program on health service utilization. Methods This was a retrospective cohort study of five assertive community treatment (ACT) teams in Indiana that implemented illness management and recovery. With Medicaid claims data from July 1, 2003, to June 30, 2008, panel data were created with person-months as the level of analysis, resulting in 14,261 observations, for a total of 498 unique individuals. Zero-inflated negative binomial regression models were used to predict hospitalization days and emergency room visits, including covariates of demographic characteristics, employment status, psychiatric diagnosis, and concurrent substance use disorder. The main predictor variables of interest were receipt of illness management and recovery services, dropout from the program, and program graduation status. Results Consumers who received some illness management and recovery services had fewer hospitalization days than those receiving only ACT. Graduates had fewer emergency room visits than did ACT-only consumers. Conclusions This is the first study to examine the impact of illness management and recovery on service utilization. Controlling for a number of background variables, the study showed that illness management and recovery programs were associated with reduced inpatient hospitalization and emergency room use over and above ACT.

show abstract

“…To deal with these situations, a hurdle model (Mullahy 1986;Zeileis et al 2008) was used. In the hurdle model, there are two component models: a truncated count model that is used for the positive counts and a hurdle component that models zero versus positive counts.…”

Section: Methodsmentioning

confidence: 99%

The relationship between fire behaviour measures and community loss: an exploratory analysis for developing a bushfire severity scale

et al. 2012

View full text Add to dashboard Cite

Current fire danger scales do not adequately reflect the potential destructive force of a bushfire in Australia and, therefore, do not provide fire prone communities with an adequate warning for the potential loss of human life and property. To determine options for developing a bushfire severity scale based on community impact and whether a link exists between the energy release rate (power) of a fire and community loss, this paper reviewed observations of 79 wildfires (from 1939 to 2009) across Victoria and other southern states of Australia. A methodology for estimating fire power based on fuel loading, fire size and progression rate is presented. McArthur's existing fire danger indices (FDIs) as well as fuel-and slope-adjusted FDIs were calculated using fire weather data. Analysis of possible relationships between fire power, FDIs, rate of spread and Byram's fireline intensity and community loss was performed using exposure as a covariate. Preliminary results showed that a stronger relationship exists between community loss and the power of the fire than between loss and FDI, although fuel-adjusted FDI was also a good predictor of loss. The database developed for this study and the relationships established are essential for undertaking future studies that require observations of past fire behaviour and losses and also to form the basis of developing a new severity scale.

show abstract

Regression Models for Count Data inR

Cited by 1,948 publications

References 19 publications

Farmers' Preferential Choice Decisions to Alternative Cassava Value Chain Strands in Morogoro Rural District, Tanzania

Farmers' Preferential Choice Decisions to Alternative Cassava Value Chain Strands in Morogoro Rural District, Tanzania

Impact of Illness Management and Recovery Programs on Hospital and Emergency Room Use by Medicaid Enrollees

The relationship between fire behaviour measures and community loss: an exploratory analysis for developing a bushfire severity scale

Contact Info

Product

Resources

About