Reduce code duplication

[echasnovski]

This is a PR for #173. It contains updates suggested in issue with major refactor of visualize(). It also has the following user-visible changes:

• Now all shading of p-value tails and intervals is done with the same transparency. In current 'master' version transparency is fixed for method = "theoretical", but varies for method = "simulation" or method = "both". This is because of actually different implementations of shading. I didn't consult if this is an intentional behaviour, but it seems unlikely to me. To see an effect, run the following code:

library(infer)
plot_iris_boot <- function(reps) {
  set.seed(101)
  
  iris_boot <- iris %>%
    tibble::as_tibble() %>%
    dplyr::mutate(
      Sepal.Length.Group = dplyr::if_else(Sepal.Length > 5, ">5", "<=5"),
      Sepal.Width.Group = dplyr::if_else(Sepal.Width > 3, "large", "small")
    ) %>%
    specify(Sepal.Length ~ Species) %>%
    hypothesize(null = "independence") %>%
    generate(reps = reps, type = "permute") %>%
    calculate(stat = "F") %>%
    visualize(method = "both", obs_stat = 2, direction = "two_sided") %>%
    print()
}

plot_iris_boot(20)
plot_iris_boot(100)

• Now visualize() works for "One sample t" theoretical distribution type for method = "both". It didn't work because it was missing in this if clause. After wrapping these checks into a function, this became apparent. To see that it wasn't working, run the following code:

library(infer)
iris %>%
  tibble::as_tibble() %>%
  dplyr::mutate(
    Sepal.Length.Group = dplyr::if_else(Sepal.Length > 5, ">5", "<=5"),
    Sepal.Width.Group = dplyr::if_else(Sepal.Width > 3, "large", "small")
  ) %>%
  specify(Petal.Width ~ NULL) %>%
  hypothesize(null = "point", mu = 1) %>%
  generate(reps = 100) %>%
  calculate(stat = "t") %>%
  visualize(method = "both")

[echasnovski]

Oh, and also in warnings about "Chi-Square" not being good for direction other than "greater" or "right" distribution is named "Chi-Square" instead of "Chi-square" for consistency.

[echasnovski]

Travis error seems to not be code fault (locally and on other R versions it passes all checks). Error comes from pandoc:
The command "curl -fLo /tmp/pandoc-2.2-1-amd64.deb https://github.com/jgm/pandoc/releases/download/2.2/pandoc-2.2-1-amd64.deb" failed and exited with 7 during .

[codecov-io]

Codecov Report

Merging #177 into develop will not change coverage.
The diff coverage is 100%.

@@           Coverage Diff           @@
##           develop   #177    +/-   ##
=======================================
  Coverage      100%   100%            
=======================================
  Files           12     12            
  Lines         1290   1135   -155     
=======================================
- Hits          1290   1135   -155

Impacted Files	Coverage Δ
R/p_value.R	100% <100%> (ø)	⬆️
R/set_params.R	100% <100%> (ø)	⬆️
R/visualize.R	100% <100%> (ø)	⬆️
R/calculate.R	100% <100%> (ø)	⬆️
R/generate.R	100% <100%> (ø)	⬆️
R/specify.R	100% <100%> (ø)	⬆️
R/utils.R	100% <100%> (ø)	⬆️

[ismayc]

Great stuff! Thanks much!

[ismayc]

@echasnovski I'm stepping away from the computer now so I'm not able to merge this. I just merged in a way to add visualise() as an alias and I suspect that's the reason for this problem. Can you resolve the conflicts so I can merge?

[github-actions]

This pull request has been automatically locked. If you believe you have found a related problem, please file a new issue (with a reprex: https://reprex.tidyverse.org) and link to this issue.