Configuration Reference

Use YAML config files for repeatable analysis runs.

Core fields

Field Required Default Notes
project.name No marketbayesmeta_analysis Human-readable project name.
project.analyst No null Analyst or owner.
project.notes No "" Short project note.
data.tracker Yes None CSV tracker path, resolved relative to the config file.
data.include_partial No false Include analysis_ready=partial rows.
analysis.evidence_type Yes None Currently pooled by default: geo_test, bls.
analysis.metric Yes None Exact metric name to pool.
analysis.scale Yes None log_relative or percentage_point.
analysis.max_abs_percent_uplift No 90 Hard limit for % geo uplift to log-relative conversion.
model.draws No 2000 Posterior draws per chain.
model.tune No 1000 Tuning draws per chain.
model.chains No 4 Number of chains.
model.target_accept No 0.9 Passed to PyMC sampling.
model.random_seed No null Sampling seed.
model.progressbar No false Keep false for scripted runs.
priors.mu_sd No scale-aware Prior SD for pooled mean mu.
priors.tau_scale No scale-aware HalfNormal scale for heterogeneity tau.
diagnostics.allow_directional No false Allow directional readiness runs only for exploratory work.
diagnostics.allow_not_recommended No false Override only for explicit exploratory work.
diagnostics.fail_on_diagnostic_failure No true Make failed sampler diagnostics blocking.
outputs.directory No output/analysis Output folder, resolved relative to config file.

Uncertainty fields

Rows with intervals must state ci_type. They must also state ci_level unless uncertainty.default_ci_level is explicitly set in config.

uncertainty:
  default_ci_level: null
  scenario_standard_errors:
    log_relative:
      low: 0.03
      medium: 0.08
      high: 0.15
    percentage_point:
      low: 1.0
      medium: 3.0
      high: 5.0

P-values are converted only when p_value_sidedness=two_sided.

Sensitivity fields

sensitivity:
  uncertainty: false
  prior: false

Custom prior sensitivity specs can be provided as sensitivity.prior_specs; when omitted the package uses scale-aware defaults.