While translating the book to the Portuguese version, we have found some problems with chunk 19: nb_reg in Time series chapter.
Line 590
observed <- predict(fitted_model, simulate_pi = FALSE)
returned an error about the tsible object, suggesting the use of bind_cols... (which dint help me a lot), but after some research and some trials, I corrected the error in the Portuguese version by transforming the "counts" object to a simple data.frame (not a tibble) prior to fitting the model, by creating auxiliary objects (count_f) to fit the models. That is, I dit it basically every time the model was fitted when there was a Fourier term.
I don't know if it was the best solution... but as far as my research on the matter went, it was the solution I could find that you cause less changes to the original code....
While translating the book to the Portuguese version, we have found some problems with chunk 19: nb_reg in Time series chapter.
Line 590 observed <- predict(fitted_model, simulate_pi = FALSE)
returned an error about the tsible object, suggesting the use of bind_cols... (which dint help me a lot), but after some research and some trials, I corrected the error in the Portuguese version by transforming the "counts" object to a simple data.frame (not a tibble) prior to fitting the model, by creating auxiliary objects (count_f) to fit the models. That is, I dit it basically every time the model was fitted when there was a Fourier term.
I don't know if it was the best solution... but as far as my research on the matter went, it was the solution I could find that you cause less changes to the original code....