2020/07/05 by Farzali Izadi, Izadi, Farzali
Mathematics · #Applications (stat.AP) #COVID-19 epidemiological studies #FOS: Biological sciences #FOS: Computer and information sciences #FOS: Physical sciences #Physics and Society (physics.soc-ph) #Populations and Evolution (q-bio.PE) #Statistical Methods and Inference
paper · pdf · doi:10.48550/arxiv.2008.01030
openalex publication_date 2020/07/05 · openalex created_date 2022/10/03 · openalex updated_date 2026/07/28
To capture the death rates and strong weekly, biweekly and probably monthly\npatterns in the Canada COVID-19, we utilize the generalized additive models in\nthe absence of direct statistically based measurement of infection rates. By\nexamining the death rates of Canada in general and Quebec, Ontario and Alberta\nin particular, one can easily figured out that there are substantial\noverdispersion relative to the Poisson so that the negative binomial\ndistribution is an appropriate choice for the analysis. Generalized additive\nmodels (GAMs) are one of the main modeling tools for data analysis. GAMs can\nefficiently combine different types of fixed, random and smooth terms in the\nlinear predictor of a regression model to account for different types of\neffects. GAMs are a semi-parametric extension of the generalized linear models\n(GLMs), used often for the case when there is no a priori reason for choosing a\nparticular response function such as linear, quadratic, etc. and need the data\nto 'speak for themselves'. GAMs do this via the smoothing functions and take\neach predictor variable in the model and separate it into sections delimited by\n'knots', and then fit polynomial functions to each section separately, with the\nconstraint that there are no links at the knots - second derivatives of the\nseparate functions are equal at the knots.\n