The ardis from atorus-research

Add_big_n function

Header N's need to be a easily formattable value within the table data

@statasaurus can you give pseudocode of how you'd want this to work

Make layer appending more verbose

Underlying issue is that if you run add_layer for the same layer twice, you get 2 layers.

Just want to make this informative and clear for users that might not understand completely.

Total rows can't be added after the fact

This works for making totals columns

demog_ard <- adsl %>%
  tardis_table(treat_var = TRT01A, where = SAFFL == "Y") %>%
  add_total_group(group_name = "Total")


# Now let's add some variables we want. First we are going to start with AGE
demog_ard %>%
  add_layer(
    group_count(target_var = AGEGR1, by = "Age (years)")
  )

# To see how this looks we are going to use the build function to build the ARD
demog_ard %>%
  build()

But this doesn't

demog_ard <- adsl %>%
  tardis_table(treat_var = TRT01A, where = SAFFL == "Y") 


# Now let's add some variables we want. First we are going to start with AGE
demog_ard %>%
  add_layer(
    group_count(target_var = AGEGR1, by = "Age (years)")
  )

# To see how this looks we are going to use the build function to build the ARD
demog_ard %>%
  add_total_group(group_name = "Total") %>%
  build()

I will note that if you make desc layers it works fine either way

Count layer transparency of denominators used

Make it more transparent that a Big N is being used as a separate denom. Currently, the logic is:

Big N's are only used for distinct counts
If population data are set, then the Big N's are used
If not, distinct denoms are calculated from the target data

Big N Caching

The overarching idea is to be able to make Big N calculations reusable between separate ARDs, so you calculate once and use many times.

Core requirements of the feature:

Class the object so appropriate methods can be developed
Add capability to import/export the Big N objects
Likely need traceability to ensure Big N data isn't obsolete when ard is generated
Dependency metadata is stored in Big N object

Potentially add label as an alternative / addition to by in group_<type>

rename `tardis_table`

Rename the function to reflect it doesn't make tables, but just ards

tadris breaks when piped indirectly

This doesn't work

adsl <- haven::read_xpt(url("https://github.com/phuse-org/TestDataFactory/raw/main/Updated/TDF_ADaM/adsl.xpt"))

# Lets set up our ard. First we are going to define the data source and
# important variables
demog_ard <- adsl %>% 
  tardis_table(treat_var = TRT01A, where = SAFFL == "Y")


# Now let's add some variables we want. First we are going to start with AGE 
demog_ard %>% 
  add_layer(
    group_desc(target_var = AGE, by = "Age (years)") 
  ) %>% 
  build()

But this does

adsl %>% 
  tardis_table(treat_var = TRT01A, where = SAFFL == "Y") %>% 
  add_layer(
    group_desc(target_var = AGE, by = "Age (years)") 
  ) %>% 
  build()

JSON/YAML Metadata Import/Export

Method of caching and ARD metadata as YAML or JSON object to load or unload a ardis object

atorus-research / ardis Goto Github PK

ardis's People

Contributors

Stargazers

Watchers

ardis's Issues

Recommend Projects

Recommend Topics

Recommend Org