FinnishCancerRegistry / vame

Package vame

vame makes it simpler to define and make use of metadata pertaining to one more variables (e.g. a tabular dataset). It implements the VariableMetadata class, which contains the metadata. The various metadata are accessed using slot functions such as vm@var_description_get.

The VariableMetadata class is intended for storing metadata for which there is "one right way". For instance, a variable has one correct description in text. This philosophy excludes tasks such as creating a manual for a dataset, which can take many forms.

See the help page ?vame::VariableMetadata for more information. In particular see the examples.

Recommended installation

devtools::install_github(
  "FinnishCancerRegistry/vame",
  ref = readline("enter latest tag on github: ")
)

TODO

Priority: high

Type: feat

vm@vame_category_space_dt:

vm@var_set_make can be used in place of vm@var_set_value_space_eval (or maybe vm@var_set_value_space_eval can call vm@var_set_make) in the case of categorical variables when multiple columns are requested AND the necessary data is available from other value spaces. Currently in v0.3.0.16 it is necessary to have ALL dependent variables in one value_space to show the dependency in the output of at least vm@vame_category_space_dt. The downside is the potential slowdown.

Maybe there needs to be a smart system, or user-input-based system, of knowing which variable sets depend on what other variable sets. For instance the user can include column var_set_dt$dep_id_set.

Priority: low

Type: feat

vm@var_set_value_space_sample:

The issue of dependent variables appears in both sampler and maker objects. It could be argued that they should be required to have the same set of dependent variables. This would mean improving the corresponding assertion functions. Alternatively, a new column var_set_dt$dep_var_nm_set or even var_set_dt$dep_id_set could be implemented --- but currently sampler objects can also sample independently, so implmenting e.g. var_set_dt$dep_id_set would require that independent samplers can still work. Currently in 0.3.0.15 using dep_var_nm_set in sampler causes data to be asserted to contain such variables. Maybe an independent sampler would need to be marked in a special way so that the data assertion is not performed.