Fig. 8From: Comparing regression modeling strategies for predicting hometimePartial dependence plots depicting the relationship between stroke severity (measured using the PaSSV score) and predicted 90-day hometime across the test data set using eight different statistical models (A Linear regression; B Ordinal logistic regression; C Poisson regression; D Negative binomial regression; E Zero-inflated Poisson regression; F Zero-inflated negative binomial regression; G Hurdle regression (negative binomial zero distribution, Poisson count distribution); H Cox proportional hazards model)Back to article page