Package index • hvtiRutilities

Data Import

Read clinical data files as stored. Type conversion is opt-in via convert_types, because coercing a two-valued numeric to logical breaks 0/1 event and censoring flags.

read_clinical_data(): Read and prepare a clinical dataset in one step

Data Type Conversion

Automatically infer and convert column types, and read back which rule fired on which column.

r_data_types(): Automatically infer and convert data types
type_conversion_report(): Read the per-column type-conversion report

Variable Labels

Extract, look up, register and override variable labels, and declare the code-to-text mappings for coded variables

label_map(): Build a lookup map of data labels
get_label(): Look up the label for a single variable
get_labels(): Look up labels for multiple variables at once
add_labels(): Add or update labels in a label map
apply_label_overrides(): Apply label overrides from a YAML file
apply_value_labels(): Declare value labels for coded variables from a YAML file
strip_level_prefix(): Strip a leading code prefix from level text
level_map(): Report what stripping a code prefix would do to a dataset's levels
clean_labels(): Apply label overrides from a YAML file (deprecated)

Data Documentation

Build data dictionaries, summarise variables, and compare dataset versions

data_dictionary(): Build a data dictionary from a labeled dataset
dataset_schema(): Column-level schema of a dataset
proc_contents(): Describe a dataset's variables, in the style of SAS PROC CONTENTS
proc_means(): Summarise numeric variables, in the style of SAS PROC MEANS
proc_univariate(): Summarise and test numeric variables, in the style of SAS PROC UNIVARIATE
proc_freq(): Frequency tables, in the style of SAS PROC FREQ
followup_check(): Check recorded follow-up before a time-related analysis
compare_datasets(): Compare two versions of a dataset

Dataset Versioning

Track and verify dataset integrity with SHA-256 checksums

update_manifest(): Create or update a dataset manifest file
verify_manifest(): Verify all datasets listed in a manifest

Data Generation

Generate labeled sample and synthetic datasets

generate_survival_data(): Generate Synthetic Cardiac Surgery Survival Data
sample_data(): Generate a sample dataset for testing

SAS Macro Canonicalization

Inventory a legacy SAS macro library and resolve name collisions

sas_triage(): Triage a directory of legacy SAS macro files
sas_macro_defs(): Extract every macro definition from a SAS file
sas_macro_signature(): Parse the documentation header of a legacy SAS macro file
write_macro_manifest(): Write the canonical macro manifest
write_collision_report(): Write the macro name-collision report

SAS Parity

Compare an R result against its SAS reference under derived tolerances

parity_tolerance(): Tolerance classes for SAS parity comparison
compare_parity(): Compare one R quantity against its SAS reference
parity_headline(): Summarise a parity table as one reviewer-facing claim

Study Setup

Initialize a study for reproducible analysis, and audit whether it is ready

study_setup(): Set up study identity and directories
study_update_identity(): Record a verified study identity
study_declare(): Declare the packages a study needs
study_need(): Install a package a study needs, or say how
register_data(): Register a study dataset
study_status(): Audit a study's reproducibility readiness
study_checklist(): Render a study audit as a markdown checklist
preflight_report(): Audit the analysis environment before running an analysis

Study Checkpoints

Snapshot a study’s code at each checkpoint, record its documents by checksum, and record its closure

study_checkpoint(): Record a study checkpoint
study_checkpoint_push(): Push pending study checkpoints
study_close() study_reopen(): Close or reopen a study

Study Manifest and Data Contract

Locate a study from its _study.yml, resolve its paths, and read and reconcile its built dataset

study_config(): Read the study manifest
study_abbreviations(): The abbreviation list a study's labels use
add_abbreviation(): Add a phrase to a study's abbreviation list
study_root(): Locate the study root
sas_path(): Build a path under the study root
built_path(): Path to the study's built dataset
built_manifest(): Record the state of the built dataset
study_dir(): Resolve a directory in the study layout
read_built(): Read the study's built dataset
review_built(): Review a rebuilt study dataset before registering it
check_data_updates(): Check a study for published dataset updates
review_data_update(): Review one published dataset release
adopt_data_update(): Adopt one published dataset release
cohort_counts(): Count the analysable cohort
assert_cohort(): Assert the cohort matches a job-level expectation

Provenance

Record what produced a filed result, in a JSON sidecar beside it

provenance_data(): Snapshot a registered data file for provenance
provenance_artifact(): Snapshot an input artifact for provenance
capture_provenance(): Capture an explicit job provenance payload
publish_provenance(): Publish captured provenance beside a completed output
record_provenance(): Capture and publish provenance for an existing output
provenance_path(): Name the provenance sidecar for an output

Caching

Compute expensive results once and reload them only while their inputs are unchanged

cache_fit(): Cache an expensive computation, strictly

Analysis Taxonomy

The two-letter prefix system inherited from the CORR analysis binder

hvti_taxonomy(): Analysis prefix taxonomy
hvti_non_prefixes(): Second fields that are not analysis prefixes
hvti_prefix_folds(): Legacy prefixes folded into a taxonomy prefix

Corpus Inventory

Sweep the job corpus by filename – which studies ran which job types

job_files(): Inventory the job files under one or more corpus roots
job_census() print(<hvti_job_census>): Roll a job-file inventory up to studies and prefixes

Candidate Pool Preparation

Getting a covariate list out of a SAS job and into a model without changing the model on the way. Each of these exists because of a failure that was silent: a comment swallowing a variable name, a mean-imputed binary being dummy-coded, a screen spending its step budget on five forms of one concept, a rank-deficient design from two names for the same thing.

sas_variable_block(): Read a variable list out of a SAS job
covariate_audit(): Audit what would happen to each covariate in a model
imputed_levels(): Levels of a column that are not clean integers
covariates_to_numeric(): Apply what covariate_audit() reports
pool_collinear_pairs(): Candidate pairs carrying the same information under unrelated names
concept_of(): The concept a candidate variable belongs to
concept_map(): Map a candidate pool to concepts
prune_to_one_form(): Prune a candidate pool to one form per concept
selection_crowding(): Concepts represented more than once among selected covariates
POOL_AFFIXES POOL_PLAIN_SUFFIX POOL_MIN_STEM: Affix conventions for grouping a SAS candidate pool into concepts

Package Utilities

hvtiRutilities-package: hvtiRutilities: Utilities to work with SAS data in R.
hvtiRutilities.news(): Display the NEWS file
r_dir_impurities(): Report top-level executable code in an R directory