Open rpruim opened 6 years ago
@dtkaplan, how does this look?
example(mod_effect) ## ## md_ffc> mod1 <- lm(wage ~ age * sex * educ + sector, data = mosaicData::CPS85) ## ## md_ffc> mod_effect(mod1, ~ sex) ## # A tibble: 1 x 6 ## change sex to_sex age educ sector ## <dbl> <fct> <fct> <dbl> <dbl> <fct> ## 1 -1.88 M F 35 12 prof ## ## md_ffc> mod_effect(mod1, ~ sector) ## # A tibble: 1 x 6 ## change sector to_sector age sex educ ## <dbl> <fct> <fct> <dbl> <fct> <dbl> ## 1 -2.23 prof clerical 35 M 12 ## ## md_ffc> mod_effect(mod1, ~ age, sex = "M", educ = c(10, 12, 16), age = c(30, 40)) ## # A tibble: 6 x 7 ## change slope age to_age sex educ sector ## <dbl> <dbl> <dbl> <dbl> <fct> <dbl> <fct> ## 1 1.18 0.118 30 40 M 10 prof ## 2 1.18 0.118 40 50 M 10 prof ## 3 1.38 0.138 30 40 M 12 prof ## 4 1.38 0.138 40 50 M 12 prof ## 5 1.78 0.178 30 40 M 16 prof ## 6 1.78 0.178 40 50 M 16 prof ## ## md_ffc> mod_effect(mod1, ~ age, sex = "F", age = 34, step = 1) ## # A tibble: 1 x 7 ## change slope age to_age sex educ sector ## <dbl> <dbl> <dbl> <dbl> <fct> <dbl> <fct> ## 1 0.0549 0.0549 34 35 F 12 prof ## ## md_ffc> mod_effect(mod1, ~ sex, age = 35, sex = "M", to = "F" ) ## # A tibble: 1 x 6 ## change sex to_sex age educ sector ## <dbl> <fct> <fct> <dbl> <dbl> <fct> ## 1 -1.88 M F 35 12 prof ## ## md_ffc> # For classifiers, the change in *probability* of a level is reported. ## md_ffc> mod2 <- rpart::rpart(sector ~ age + sex + educ + wage, data = mosaicData::CPS85) ## ## md_ffc> mod_effect(mod2, ~ educ) ## # A tibble: 1 x 7 ## change_clerical slope educ to_educ age sex wage ## <dbl> <dbl> <dbl> <dbl> <dbl> <fct> <dbl> ## 1 0 0 12 14 35 M 7.8 ## ## md_ffc> mod_effect(mod2, ~ educ, class_level = "manag") ## # A tibble: 1 x 7 ## change_manag slope educ to_educ age sex wage ## <dbl> <dbl> <dbl> <dbl> <dbl> <fct> <dbl> ## 1 0 0 12 14 35 M 7.8
I see that sometimes the change column is inheriting a variable name. Do we want that? Should it be there in all cases?
@dtkaplan, how does this look?