rafalab / excessmort

R package for estimating excess deaths from daily count data

Hi there 👋

• I am a Professor and Chair of the Department of Data Science at Dana-Farber Cancer Institute.
• Also Professor of Applied Statistics at Harvard.
• My research focuses on Genomics but I am interested in applications in general.
• I teaches several Data Science courses.
• I blog posts at Simply Statistics.
• The best way to keep up with the latest news is to follow @rafalab.

Learn more about me at http://rafalab.dfci.harvard.edu/

excessmort's Issues

No s.e. from the fitted model of expected death

Dear Professor Irizarry,

I found that as “expected deaths” were calculated, the package itself only gave the estimate of expected without standard error. In this case, -compute_expected- and -expected_plot- do not give the confidence intervals of the estimates.

I would then like to suggest you to integrate standard error so as to calculate confidence interval for the estimates.

Your package is nicely written. Thank you so much for your awesome work!

Best,
Enoch

Error with monthly data

I'm trying this package out on monthly mortality data in Georgia. I've got the following variables:

• Date (year, month, day--but just the first date of each month)
• Outcome (count of deaths)
• Population

There's 147 observations going back to 2000, and I exclude dates from 2020. When I run expected_count, I get the following error:

Error: Column expected must be length 7 (the group size) or one, not 147

Here's the function:
expect_test<-compute_expected(data_test,
frequency=7,
exclude=excludedates)

It looks like it's trying to created 1 expected count for each month but there's 20 years of data. Am I somehow setting this up wrong?

Covariates

Is there a way to include covariates such as temperature?

expected_plot() returns an error when compute_expected(..., keep.components = TRUE).

Not sure if this is expected behavior, but when using expected_plot(), I will get an error (Error in if (!attr(expected, "frequency") %in% c(365, 52)) warning("This function assumes weekly or daily data. This dataset has ", : argument is of length zero) if the object comes from compute_expected(..., keep.components = TRUE) but not if keep.components = FALSE despite all other parameters (and data) being identical.

If expected behavior, I think a more descriptive error would be helpful?