predictor_geography_scope
Back to L1 | Browse all axes | Browse all options
Axis
predictor_geography_scopeon sub-layerl1_d(layerl1).
Sub-layer
l1_d
Axis metadata
Default:
'match_target'Sweepable: False
Status: operational
Operational status summary
Operational: 4 option(s)
Future: 0 option(s)
Options
match_target – operational
Use the same geography scope as the target.
Default; predictor states match the L1.D target_geography_scope. Ensures spatial coherence for state-level forecasts.
When to use
Default for state-level forecasts.
References
macroforecast design Part 1, L1: ‘data definition is the recipe layer that pins source, target, geography, and horizon – everything downstream branches off these choices.’
Related options: all_states, selected_states, national_only
Last reviewed 2026-05-05 by macroforecast author.
all_states – operational
Use predictors from every state regardless of target geography.
All-50-states predictor block. Useful when cross-state spillovers matter and the target is a single state.
When to use
Spillover / cross-state interaction studies.
References
macroforecast design Part 1, L1: ‘data definition is the recipe layer that pins source, target, geography, and horizon – everything downstream branches off these choices.’
Related options: match_target, selected_states
Last reviewed 2026-05-05 by macroforecast author.
selected_states – operational
Use predictors from a user-supplied state list.
Reads leaf_config.predictor_states and restricts the predictor block to that subset.
When to use
Custom regional studies (e.g. neighbouring states).
References
macroforecast design Part 1, L1: ‘data definition is the recipe layer that pins source, target, geography, and horizon – everything downstream branches off these choices.’
Related options: match_target, all_states
Last reviewed 2026-05-05 by macroforecast author.
national_only – operational
Use only national-aggregate predictors.
Strips state-level predictors and keeps only national series. Reduces panel dimension when state-level features are noise.
When to use
When national variables alone explain target variation.
When NOT to use
State-level forecasts where regional predictors carry signal.
References
macroforecast design Part 1, L1: ‘data definition is the recipe layer that pins source, target, geography, and horizon – everything downstream branches off these choices.’
Related options: all_states, match_target
Last reviewed 2026-05-05 by macroforecast author.