Develop shadow_shift method for factors - perhaps add another level (smaller than smallest))

[njtierney]

No description provided.

[njtierney]

library(narnia)
library(tidyverse)
#> Loading tidyverse: ggplot2
#> Loading tidyverse: tibble
#> Loading tidyverse: tidyr
#> Loading tidyverse: readr
#> Loading tidyverse: purrr
#> Loading tidyverse: dplyr
#> Conflicts with tidy packages ----------------------------------------------
#> filter(): dplyr, stats
#> lag():    dplyr, stats

brfss %>%
  add_shadow_shift("STOPSMK2") %>%
  select(STOPSMK2, STOPSMK2_shift)
#> # A tibble: 245 x 2
#>    STOPSMK2 STOPSMK2_shift
#>      <fctr>         <fctr>
#>  1       NA        missing
#>  2       NA        missing
#>  3       NA        missing
#>  4       NA        missing
#>  5      Yes            Yes
#>  6       NA        missing
#>  7       NA        missing
#>  8       NA        missing
#>  9      Yes            Yes
#> 10       NA        missing
#> # ... with 235 more rows

Need to think about:

• visualisations for this
• standardised way to code this value in a factor - it should be one smaller than the smallest value

[njtierney]

I think that I am OK with this producing values that are larger than "present"

library(tidyverse)
#> Loading tidyverse: ggplot2
#> Loading tidyverse: tibble
#> Loading tidyverse: tidyr
#> Loading tidyverse: readr
#> Loading tidyverse: purrr
#> Loading tidyverse: dplyr
#> Conflicts with tidy packages ----------------------------------------------
#> filter(): dplyr, stats
#> lag():    dplyr, stats
library(narnia)

brfss %>%
  add_shadow_shift("STOPSMK2") %>%
  select(STOPSMK2, STOPSMK2_shift) %>%
  mutate(smk_lvl = as.numeric(STOPSMK2),
         smk_lvl_2 = as.numeric(STOPSMK2_shift))
#> # A tibble: 245 x 4
#>    STOPSMK2 STOPSMK2_shift smk_lvl smk_lvl_2
#>      <fctr>         <fctr>   <dbl>     <dbl>
#>  1       NA        missing      NA         3
#>  2       NA        missing      NA         3
#>  3       NA        missing      NA         3
#>  4       NA        missing      NA         3
#>  5      Yes            Yes       1         1
#>  6       NA        missing      NA         3
#>  7       NA        missing      NA         3
#>  8       NA        missing      NA         3
#>  9      Yes            Yes       1         1
#> 10       NA        missing      NA         3
#> # ... with 235 more rows

[njtierney]

Force NA category to be higher value for factors

[njtierney]

Factors with NA get plotted, I currently don't see a problem with the current behaviour of shadow_shift.factor -

library(naniar)
library(tidyverse)
#> ── Attaching packages ────────────────────────────────────────────── tidyverse 1.2.1 ──
#> ✔ ggplot2 2.2.1     ✔ purrr   0.2.4
#> ✔ tibble  1.4.1     ✔ dplyr   0.7.4
#> ✔ tidyr   0.7.2     ✔ stringr 1.2.0
#> ✔ readr   1.1.1     ✔ forcats 0.2.0
#> ── Conflicts ───────────────────────────────────────────────── tidyverse_conflicts() ──
#> ✖ dplyr::filter() masks stats::filter()
#> ✖ dplyr::lag()    masks stats::lag()

riskfactors %>% 
  cast_shadow(smoke_days) %>%
  add_shadow_shift(smoke_days)
#> # A tibble: 245 x 3
#>    smoke_days smoke_days_NA smoke_days_shift
#>    <fct>      <fct>         <fct>           
#>  1 <NA>       NA            missing         
#>  2 <NA>       NA            missing         
#>  3 <NA>       NA            missing         
#>  4 <NA>       NA            missing         
#>  5 Everyday   !NA           Everyday        
#>  6 <NA>       NA            missing         
#>  7 <NA>       NA            missing         
#>  8 Not@All    !NA           Not@All         
#>  9 Everyday   !NA           Everyday        
#> 10 Not@All    !NA           Not@All         
#> # ... with 235 more rows

riskfactors %>%
  ggplot(aes(x = smoke_days,
             y = bmi)) + 
  geom_point()
#> Warning: Removed 11 rows containing missing values (geom_point).

riskfactors %>%
  ggplot(aes(x = smoke_days,
             y = bmi)) + 
  geom_miss_point()

riskfactors %>%
  ggplot(aes(x = smoke_days,
             y = activity_limited)) + 
  geom_point()

riskfactors %>%
  ggplot(aes(x = smoke_days,
             y = activity_limited)) + 
  geom_miss_point()

I'm closing this for the moment, but I will reopen it if need be (any thoughts @dicook?)