vision

Overview

vision provides spec-first workflow infrastructure for consumer psychology research. It parses and validates YAML or JSON specification files, constructs auditable analysis recipes, and enforces schema compliance. vision does not perform data ingestion, modeling, or reporting—it focuses solely on schema validation and recipe construction to ensure reproducible, audit-ready research workflows.

vision is part of the niche R universe for reproducible consumer research.

Installation

You can install the development version of vision from GitHub:

# install.packages("devtools")
devtools::install_github("phdemotions/vision")

Minimal Example

Here's a complete end-to-end workflow using vision:

library(vision)

# 1. Create a specification template
template_path <- tempfile(fileext = ".yml")
write_spec_template(template_path)

# 2. Read the specification
spec <- read_spec(template_path)

# 3. Validate the specification against schema v0.1.0
spec <- validate_spec(spec)

# 4. Build an auditable recipe
recipe <- build_recipe(spec)

# 5. Write the recipe to disk
write_recipe(recipe)

# The recipe is now saved as JSON in the audit directory
# and can be used by downstream packages for analysis execution

Key Functions

• schema_version_current(): Returns the current schema version ("0.1.0")
• read_spec(path): Parse YAML/JSON specification files
• validate_spec(spec): Validate against schema v0.1.0 requirements
• build_recipe(spec): Construct auditable analysis recipe with deterministic hashing
• write_recipe(recipe, path): Serialize recipe to canonical JSON
• write_spec_template(path): Generate a minimal compliant YAML template

What vision Does NOT Do

vision is intentionally minimal and focused:

• ❌ Does not read or validate data files
• ❌ Does not score scales or compute composites
• ❌ Does not run statistical models
• ❌ Does not produce tables, figures, or reports

For complete analysis workflows, use vision in combination with other packages in the niche universe.

Schema v0.1.0

vision validates specifications against schema version 0.1.0, which requires:

Top-level fields:

• meta: Study metadata (title, authors, created date)
• data: Data sources and ID column specification
• vars: Variable renaming and type specifications
• rules: Exclusion and missingness handling rules
• scales: Scale definitions with items and compositing methods
• models: Statistical model specifications
• outputs: Output directory configuration

See the vignette for detailed schema requirements and examples.

License

MIT