A parks and recreation demand desk wants reproducibility for a promoted attendance model. What should they store together as one artifact set?
Select an answer to reveal the explanation.
Short Explanation
Bundle the model binary, config YAML, and metrics JSON like a go-bag—repro needs the whole set, not a flyer.
Full Explanation
A parks and recreation demand desk that wants reproducibility for a promoted attendance model should store the model binary, configuration YAML, and metric JSON together as one artifact set. Reproducibility requires the weights (or serialized estimator), the configuration that produced them, and the recorded metrics that justified promotion. Bundling those pieces lets another analyst reload the same champion, verify reported scores, and audit parks demand decisions months later.
Storing only a verbal description of the model with no files fails by concept: oral summaries cannot be reloaded or verified computationally. Keeping only a marketing flyer about the parks season fails because promotional material is not a model artifact set. Relying only on an unlocked spreadsheet with ad-hoc formulas fails because spreadsheets lack a governed binary-plus-config-plus-metrics package and drift easily.
Exam caveat: reproducibility is a package problem—binary, config, and metrics—not a single file or a narrative substitute.
Operational check: on promotion, write one versioned bundle containing the attendance model binary, YAML config (including feature list and training settings), and metrics JSON; restore that bundle in a clean environment and confirm metrics regenerate within tolerance.