Unify the `epi_lag` and `epi_ahead` functions

[dajmcdon]

These are basically the same. Try to remove as much extraneous repetition as possible (say by dispatching to common underlying functions).

[kenmawer]

Currently on my fork, I'm having an error here. I wilk continue work on it tomorrow.

e

[dajmcdon]

I cannot reproduce your error locally. Can you describe how you are running the script? What R/package versions you're using?

[kenmawer]

I cannot reproduce your error locally. Can you describe how you are running the script? What R/package versions you're using?

On the example recipe, I run this with Ctrl+Shift+R (can also do Ctrl+A then Ctrl+R).

R version 4.1.1

tidyverse 1.3.1

covidcast 0.4.2

delphi.epidata 1.0.0

epiprocess 1.0.0

tidymodels 0.2.0

Note that I am getting the error with the code shown on my kenmawer/epipredict. I haven't tried it out with cmu-delphi/epipredict yet despite it differing from mine due to the inclusion of epi_shift.

It is saved in step_epi_lag (will be refactored to step_epi_shift once the code runs properly).

[dajmcdon]

Have you loaded epipredict? (with devtools::load_all()) Note: the function epi_shift() is never called by that example script. (It is used in other, unrelated parts of the package.)

[kenmawer]

Have you loaded epipredict? (with devtools::load_all()) Note: the function epi_shift() is never called by that example script. (It is used in other, unrelated parts of the package.)

Yes, I have. I think we should discuss this issue in detail as I followed all your steps but I still get this error, implying I may have done something else wrong.

[kenmawer]

I finished this. I am suggesting there be something done about insertion of negative values (which I utilized for tests), such as a warning.

[kenmawer]

Just a note: I still need to work on improving documentation for this, as the documentation is not quite right, which I didn't quite notice until recently.