Skip to main content
Ctrl+K

macroforecast

  • Getting started
  • User guide
  • Reference
  • Replications
  • Help
  • GitHub
  • PyPI
  • Getting started
  • User guide
  • Reference
  • Replications
  • Help
  • GitHub
  • PyPI

Section Navigation

  • Encyclopedia
    • Layer L0 – Study Setup
      • failure_policy
      • reproducibility_mode
      • compute_mode
    • Layer L1 – Data
      • custom_source_policy
      • dataset
      • frequency
      • information_set_type
      • vintage_policy
      • fred_sd_frequency_policy
      • target_structure
      • variable_universe
      • missing_availability
      • raw_missing_policy
      • raw_outlier_policy
      • release_lag_rule
      • contemporaneous_x_rule
      • official_transform_policy
      • official_transform_scope
      • target_geography_scope
      • predictor_geography_scope
      • fred_sd_state_group
      • state_selection
      • fred_sd_variable_group
      • sd_variable_selection
      • sample_start_rule
      • sample_end_rule
      • horizon_set
      • regime_definition
      • regime_estimation_temporal_rule
    • Layer L1.5 – Data summary
      • coverage_view
      • summary_metrics
      • summary_split
      • stationarity_test
      • stationarity_test_scope
      • missing_view
      • outlier_view
      • correlation_method
      • correlation_view
      • diagnostic_format
      • attach_to_manifest
      • figure_dpi
      • latex_export
    • Layer L2 – Preprocessing
      • sd_series_frequency_filter
      • mixed_frequency_representation
      • quarterly_to_monthly_rule
      • monthly_to_quarterly_rule
      • transform_policy
      • sd_tcode_policy
      • transform_scope
      • outlier_policy
      • outlier_action
      • outlier_scope
      • imputation_policy
      • imputation_temporal_rule
      • imputation_scope
      • frame_edge_policy
      • frame_edge_scope
    • Layer L2.5 – Pre vs post preprocessing
      • comparison_pair
      • comparison_output_form
      • distribution_metric
      • distribution_view
      • correlation_shift
      • correlation_method
      • cleaning_summary_view
      • t_code_application_log
      • diagnostic_format
      • attach_to_manifest
      • figure_dpi
      • latex_export
    • Layer L3 – Feature engineering
      • op
    • Layer L3.5 – Feature diagnostics
      • comparison_stages
      • comparison_output_form
      • factor_view
      • dfm_diagnostics
      • feature_correlation
      • correlation_method
      • correlation_view
      • lag_view
      • marx_view
      • selection_view
      • stability_metric
      • diagnostic_format
      • attach_to_manifest
      • figure_dpi
      • latex_export
    • Layer L4 – Forecasting model
      • family
      • forecast_strategy
      • training_start_rule
      • refit_policy
      • search_algorithm
      • pi_correction
    • Layer L4.5 – Generator diagnostics
      • fit_view
      • fit_per_origin
      • forecast_scale_view
      • back_transform_method
      • window_view
      • coef_view_models
      • tuning_view
      • ensemble_view
      • weights_over_time_method
      • diagnostic_format
      • attach_to_manifest
      • figure_dpi
      • latex_export
    • Layer L5 – Evaluation
      • primary_metric
      • point_metrics
      • density_metrics
      • direction_metrics
      • relative_metrics
      • benchmark_window
      • benchmark_scope
      • agg_time
      • agg_horizon
      • agg_target
      • agg_state
      • oos_period
      • regime_use
      • regime_metrics
      • decomposition_target
      • decomposition_order
      • ranking
      • report_style
    • Layer L6 – Statistical tests
      • equal_predictive_test
      • loss_function
      • model_pair_strategy
      • hln_correction
      • nested_test
      • nested_pair_strategy
      • cw_adjustment
      • enc_test_one_sided
      • cpa_test
      • cpa_window_type
      • cpa_conditioning_info
      • cpa_critical_value_method
      • multiple_model_test
      • mcs_alpha
      • mmt_loss_function
      • bootstrap_method
      • bootstrap_n_replications
      • bootstrap_block_length
      • mcs_t_statistic
      • spa_studentization
      • stepm_alpha
      • density_test
      • interval_test
      • coverage_levels
      • pit_n_bins
      • pit_test_horizon_dependence
      • direction_test
      • direction_threshold
      • direction_alpha
      • residual_test
      • residual_lag_count
      • residual_test_scope
      • residual_alpha
    • Layer L7 – Interpretation / importance
      • output_table_format
      • figure_type
      • top_k_features_to_show
      • precision_digits
      • figure_dpi
      • figure_format
      • latex_table_export
      • markdown_table_export
      • op
    • Layer L8 – Output / provenance
      • export_format
      • compression
      • saved_objects
      • model_artifacts_format
      • provenance_fields
      • manifest_format
      • artifact_granularity
      • naming_convention
    • Browse by layer
    • Browse by axis
    • Browse by option
    • Public Python API
  • Architecture
    • Layer 0: Study Setup
      • Layer 0 Axis: study_scope
      • Layer 0 Axis: failure_policy
      • Layer 0 Axis: reproducibility_mode
      • Layer 0 Axis: compute_mode
      • Registry Catalog: axis_type
    • Layer 1: Data Source, Target y, Predictor x
      • Data Source Mode / Frequency
      • Forecast-Time Information
      • Target (y) And Predictor (x) Definitions
      • FRED-SD Predictor Scope
      • Raw Source Cleaning
      • Official Transforms
      • Frame Availability
    • Layer 2: Preprocessing
    • Layer 3: Feature Engineering
    • Layer 4: Forecasting Model
    • Layer 5: Evaluation
    • Layer 6: Statistical Tests
    • Layer 7: Interpretation / Importance
    • Layer 8: Output / Provenance
    • Foundation Core
    • Philosophy
    • Layer Boundary Contract
    • Recipe Layers
    • Artifacts And Manifest
    • Reproducibility
    • Terminology
  • Reference
  • Encyclopedia
  • Layer L4 – Forecasting model

Layer L4 – Forecasting model#

Back to encyclopedia | Browse layers | Browse all axes

  • Layer ID: l4

  • Category: construction

  • Sub-layers: 4

  • Axes: 6

  • Options across axes: 64

Sub-layers#

Sub-layer

Name

Gate

Axes

L4_A_model_selection

Model selection

always

family

L4_B_forecast_strategy

Forecast strategy

always

forecast_strategy

L4_C_training_window

Training window

always

training_start_rule, refit_policy

L4_D_tuning

Tuning

always

search_algorithm

previous

latex_export

next

family

On this page
  • Sub-layers
Edit on GitHub
Show Source

© Copyright 2026, NanyeonK.

Created using Sphinx 9.0.4.