What we’d like to implement

For future CRAN releases

  • Shading for Confidence Intervals in visualize()
    • Need to implement for theoretical distributions and method = "both"
  • Allow visualize() to specify which variable to plot instead of only working with stat column
  • Check that corresponding z for one prop depends on params being set in hypothesize with specify() %>% calculate() shortcut
  • Check that assumptions have been met for the theoretical distribution and "both"
  • Fix double printing of Response: and Explanatory:
    • Might have something to do with infer class being set in multiple spots?
  • Check that specify() %>% calculate() works for stat = "z" and stat = "t"
    • Determine if other wrapper functions should be created
      • z_test(), mean_stat(), diff_in_mean_stat(), etc.?
  • Consider re-working how p-values can be calculated (both for computational and theoretical)
    • Maybe check to make sure that p-values don’t exceed 1 too
  • Should the wrapper functions like t_test() also include a logical conf_int argument?
  • Check that stat is calculated appropriately if generate() is not called
    • order of the arguments into the standardized statistics should be checked again
    • May require moving of set_params() to be after specify()
    • Update visualize() tests to include calculate() step for method = "theoretical"
  • Add add_obs_stat toggle into visualize() and p-value calculation?
  • Shift to list-columns in generate()
  • Implement check of stat in theoretical visualize()
  • Write test to check that bootstrapped values are centered near the hypothesized value in specify()
  • Write vignettes on how NOT to use infer (strange errors, funky results, etc.)
  • Implement theoretical distributions for bootstrap distributions
  • Create resources.md with links to slides/talks/workshops given about infer
  • Add Lionel’s vis checking package (vdiffr) to visualize() tests
  • Determine whether calculate() should be where the set_params() function is called instead of in specify()
  • Need to also add parameters to wrapper functions so that randomization methods can be implemented by practitioners looking to skip the longer pipe syntax
  • Find a better word than "simulate" for type in generate()