Author: Svetoslav Bliznashki
Title: A Bayesian Logistic Growth Model for the Spread of COVID-19 in New York Document date: 2020_4_7
ID: lhv83zac_25
Snippet: Looking at figures 2 and 5 we see that the errors, in all likelihood, both lack homoscedasticity and possess an auto-correlated structure. In order to (partially) alleviate these problems we removed the normality assumption present above and replaced it with the assumption that the errors follow a t-distribution with location parameter equal to 0 and scale (similar to the standard deviation used above) and degrees of freedom (df) parameters estim.....
Document: Looking at figures 2 and 5 we see that the errors, in all likelihood, both lack homoscedasticity and possess an auto-correlated structure. In order to (partially) alleviate these problems we removed the normality assumption present above and replaced it with the assumption that the errors follow a t-distribution with location parameter equal to 0 and scale (similar to the standard deviation used above) and degrees of freedom (df) parameters estimated from the data (see Kruschke, 2012 for the same approach in the context of a linear model).
Search related documents:
Co phrase search for related documents, hyperlinks ordered by date