Add class

[zkamvar]

This PR adds the basic data dictionary where the user would set desired variables with the epivars() function and then label their column names with the as_linelist() function:

# User modifies this -------------------------
epivars("date_release", "age_months", set = TRUE)
ll <- as_linelist(dat,
  date_release = "release",
  age_months   = "months"
)

# This will run with any data set ------------
get_var(ll, "date_release")
get_var(ll, "age_months")

example workflow with linelist

library('outbreaks')
library('incidence')
library('linelist')
#> linelist is loaded with the following global variables in `epivars()`:
#> id, date_onset, date_report, gender, age, age_group, geo
# define the important variables
epivars(
        "id",
        "date_onset",
        "gender",
        "age",
        "outcome", 
        "date_outcome", 
        "date_hospitalisation", 
        set = TRUE)
#>  [1] "id"                   "date_onset"           "date_report"         
#>  [4] "gender"               "age"                  "age_group"           
#>  [7] "geo"                  "outcome"              "date_outcome"        
#> [10] "date_hospitalisation"

# convert the data set to a linelist and define the variables
ll <- as_linelist(outbreaks::fluH7N9_china_2013,
  "outcome"              = "outcome", 
  "date_outcome"         = "date_of_outcome", 
  "date_hospitalisation" = "date_of_hospitalisation",
  "id"                   = "case_id",
  "date_onset"           = "date_of_onset",
  "gender"               = "gender",
  "age"                  = "age"
  )
head(ll)
#>   case_id date_of_onset date_of_hospitalisation date_of_outcome outcome
#> 1       1    2013-02-19                    <NA>      2013-03-04   Death
#> 2       2    2013-02-27              2013-03-03      2013-03-10   Death
#> 3       3    2013-03-09              2013-03-19      2013-04-09   Death
#> 4       4    2013-03-19              2013-03-27            <NA>    <NA>
#> 5       5    2013-03-19              2013-03-30      2013-05-15 Recover
#> 6       6    2013-03-21              2013-03-28      2013-04-26   Death
#>   gender age province
#> 1      m  87 Shanghai
#> 2      m  27 Shanghai
#> 3      f  35    Anhui
#> 4      f  45  Jiangsu
#> 5      f  48  Jiangsu
#> 6      f  32  Jiangsu
head(outbreaks::fluH7N9_china_2013)
#>   case_id date_of_onset date_of_hospitalisation date_of_outcome outcome
#> 1       1    2013-02-19                    <NA>      2013-03-04   Death
#> 2       2    2013-02-27              2013-03-03      2013-03-10   Death
#> 3       3    2013-03-09              2013-03-19      2013-04-09   Death
#> 4       4    2013-03-19              2013-03-27            <NA>    <NA>
#> 5       5    2013-03-19              2013-03-30      2013-05-15 Recover
#> 6       6    2013-03-21              2013-03-28      2013-04-26   Death
#>   gender age province
#> 1      m  87 Shanghai
#> 2      m  27 Shanghai
#> 3      f  35    Anhui
#> 4      f  45  Jiangsu
#> 5      f  48  Jiangsu
#> 6      f  32  Jiangsu

get_meta(ll)
#>                    column               epivar  class hxl
#> 1                 case_id                   id factor    
#> 2           date_of_onset           date_onset   Date    
#> 3 date_of_hospitalisation date_hospitalisation   Date    
#> 4         date_of_outcome         date_outcome   Date    
#> 5                 outcome              outcome factor    
#> 6                  gender               gender factor    
#> 7                     age                  age factor    
#> 8                province                 <NA> factor

head(gender(ll))
#> [1] m m f f f f
#> Levels: f m
head(get_var(ll, "outcome"))
#> [1] Death   Death   Death   <NA>    Recover Death  
#> Levels: Death Recover
ic <- incidence(date_onset(ll), interval = "week", groups = gender(ll))
#> 10 missing observations were removed.
plot(ic)

^{Created on 2019-01-18 by the reprex package (v0.2.1)}

The roadmap I'm thinking about is this:

• basic data dictionary functionality
• A data dictionary that allows you to map standard variable names to columns
• Integration with #hxl standard for dynamic search of variables
• Validation of categorical values
• Automatically generate linelist from an imported dictionary using the openxlsx package, setting defaults for the columns.
• Import dictionary and validation from file
• explore possibility for adding a global dynamic function environment that would update based on the global variables used. For example, if the user specifies epivars("something", set = TRUE), then they can access that var from either get_var(x, "something") or get$something(x).