Day 3

Note

Set up environment and import package. Total time used: 2h 44m (6.8%)

15.33-15.53, 15.57-16.00: Browse code and set up environment

The study uses the EPIC model, with the repository forked from that package and then with code for this particular analysis using that model added - I think, all to casedetection/. I can see that casedetection/Case_Detection_Results.md has similar tables and figures to the scope (table 3, figure 3).

The README has an additional section titled Case Detection Analysis with instructions to perform the analysis. It suggests to install the branch as a package or to clone the repository, then copy the code from Case_Detection_results.Rmd and Case_Detection_Results_5yrs.Rmd, and knit it. It warns that the run time is about 16 hours for each script, and states that the analysis was conducted using R version 3.5.3.

Following the procedure as I have done for previous R projects that I am reproducing, I will initially try to do so using the latest version of R and packages, and then attempt to backdate if there are issues.

There is a DESCRIPTION file with required packages for EPIC - since it uses that model, I’m assuming this to likewise be the basis for the environment for this reproduction. I created a DESCRIPTION file with the depends, packages and suggests from that, and used it to set-up renv, so:

• renv::init(bare=TRUE)
• renv::install() - but “Error: package ‘rredis’ is not available”
  • On CRAN page can see that the package was removed from CRAN and archived on 2022-03-21, but [can access archive] (https://cran.r-project.org/src/contrib/Archive/rredis/).
  • Installed remotes - renv::install("remotes")
  • Used that to install the latest version of rredis in the CRAN archive - remotes::install_version("rredis", "1.7.0")
  • Add the installation of that version via remotes to DESCRIPTION
  • Deleted current renv (so I could test that DESCRIPTION was set up correct with remotes). So, renv::deactivate(), delete renv/, renv::init(bare=TRUE), renv::install()
• renv::snapshot()

16.01-16.41: Import epicR

Initially, I tried adding epicR to the environment, and then running the .Rmd file. To DESCRIPTION I tried:

• KateJohnson/epicR@closed_cohort
• git::https://github.com/KateJohnson/epicR.git@closed_cohort
• url::https://github.com/KateJohnson/epicR/archive/refs/heads/closed_cohort.zip
• github::KateJohnson/epicR@c239c55a700f9e97869c16d530080a354ec631568
• github::KateJohnson/epicR@c239c55a
• github::KateJohnson/epicR.git@closed_cohort

However, these did not work. The final three found the package, but then had error code 22.

I then just did manual installation following the instructions provided (remotes::install_github('KateJohnson/epicR', ref="closed_cohort")), which had error message that dependency rredis is not available. Hence, it seems I’ll need to just use it as a local package anyway.

Tried to figure the simplest way to install the package, whilst having a simpler folder layout. Figured the simplest might be to use devtools::load_all() to load a local folder as a package. I add devtools to DESCRIPTION then re-ran renv::install() and renv::snapshot().

I copied the package into a folder, but amended its DESCRIPTION file so that rredis is installed using remotes. In the .Rmd file, I then devtools::load_all("./epicR"). However, this had an error:

RWarning: ── Conflicts ───────────────────────────────────── epicR conflicts
──
✖ `Cget_agent_events` masks `epicR::Cget_agent_events()`.
✖ `Cget_all_events` masks `epicR::Cget_all_events()`.
✖ `Cget_all_events_matrix` masks `epicR::Cget_all_events_matrix()`.
  … and more.
ℹ Did you accidentally source a file rather than using `load_all()`?
  Run `rm(list = c("Cget_agent_events", "Cget_all_events",
  "Cget_all_events_matrix", "Cget_event", "Cget_events_by_type",
  "Cget_inputs", "Cget_n_events", "Cget_output", "Cget_output_ex",
  "Cget_runtime_stats", "Cget_settings", "Cget_smith",
  "Cset_input_var", "Cset_settings_var", "get_sample_output",
  "mvrnormArma"))` to remove the conflicts.

Despite that though, it seemed to have worked.

I then add some additional dependencies from the .Rmd to DESCRIPTION and to the renv.

Timings

import sys
sys.path.append('../')
from timings import calculate_times

# Minutes used prior to today
used_to_date = 101

# Times from today
times = [
    ('15.33', '15.53'),
    ('15.57', '16.00'),
    ('16.01', '16.41')]

calculate_times(used_to_date, times)

Time spent today: 63m, or 1h 3m
Total used to date: 164m, or 2h 44m
Time remaining: 2236m, or 37h 16m
Used 6.8% of 40 hours max