target_geography_scope

Back to L1 | Browse all axes | Browse all options

Axis target_geography_scope on sub-layer l1_d (layer l1).

Sub-layer

l1_d

Axis metadata

  • Default: 'all_states'

  • Sweepable: False

  • Status: operational

Operational status summary

  • Operational: 3 option(s)

  • Future: 0 option(s)

Options

single_state – operational

Single FRED-SD state target (e.g., California payrolls).

Selects one US state as the target. Requires leaf_config.target_state (two-letter postal code). Predictors default to match_target (same state).

When to use

State-level case studies (e.g., CA / TX / NY-specific 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, predictor_geography_scope

Last reviewed 2026-05-04 by macroforecast author.

all_states – operational

Forecast every state’s series jointly (50+DC targets).

Treats every state series as a target. The L5 metrics table carries one row per (model, state, horizon, origin) and the L7 us_state_choropleth figure type maps importance scores to the geographic layout.

This is the standard FRED-SD configuration for cross-state comparison studies.

When to use

Geographic-importance studies; cross-state benchmark comparisons.

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: single_state, selected_states, fred_sd_state_group

Last reviewed 2026-05-04 by macroforecast author.

selected_states – operational

Forecast a user-supplied subset of states.

Like all_states but restricted to leaf_config.target_states = [postal_codes...] or to a named fred_sd_state_group (census regions / divisions, BEA regions, etc.).

When to use

Region-specific studies (Northeast vs. Midwest), Census-division comparisons.

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, fred_sd_state_group

Last reviewed 2026-05-04 by macroforecast author.