Reference#

New here? Start with the User Guide for concept-level explanations and a getting-started walkthrough.

Current workflow reference for the clean public Python API and callable module boundaries. Dataset descriptions live in FRED Datasets.

Start with Documentation Map if you are deciding what to inspect. Use this page when you already know the part of the workflow you want to read.

First Review Path#

Order	Page	Why it comes first
1	Documentation Map	Shows which page answers which question.
2	Workflow Contract	Defines module ownership and runner composition.
3	Legacy Callable Coverage	Confirms what old runtime functionality is covered, removed, or deferred.
4	Public Python API	Lists importable public symbols.
5	Reference Verification	Shows formula/reference anchors and future verification expansion.

Group	Pages	Owns
Orientation	Documentation Map, Workflow, Legacy Coverage, Reference Verification, Public API	How the package is organized and what is currently covered.
Package Configuration	Meta	Package-wide defaults used only when a direct function or runner policy does not pass a more specific value.
Data Pipeline	Data, Preprocessing, Data Analysis	Data loading, metadata, cleaning, summaries, and before/after checks. See FRED Datasets for dataset pages.
Feature Pipeline	Filters, Feature Engineering, Feature Analysis	One-series filters/smoothers, targets, predictors, transforms, factor/selection tools, and feature diagnostics.
Forecast Pipeline	Window, Models, Model Ensemble, Model Selection, Forecasting, Forecast Analysis	Timing, model fits, fit-time model composition, parameter search, runner execution, and forecast diagnostics.
Evaluation and Testing	Metrics, Tests, Evaluation	Scores, statistical tests, benchmark comparisons, regimes, aggregation, and reports.
Explanation and Delivery	Interpretation, Dual Interpretation, Output, Reporting	Attribution, observation-based dual interpretation, output generation, artifact writing, paper-table presets, and report/table rendering.
Orchestration	Pipeline	Comprehensive POOS engine: arms, t-code target resolution, relative RMSE + DM/CW/MCS, combinations, deferred interpretation, and a single report.
Custom Hooks	Custom Extensions	User-provided datasets, preprocessing, features, models, policies, tests, interpretation, and artifacts.

If you want to check…	Open first	Then open
Current package structure	Documentation Map	Workflow Contract, Public Python API
Package-wide defaults	Meta	Forecasting for runner policy usage.
Whether old runtime features are covered	Legacy Callable Coverage	Reference Verification
Data and preprocessing behavior	Data	FRED Datasets, Preprocessing
Paper replication settings	Replication Gallery	Reference Verification, Forecasting
Forecast runner behavior	Workflow Contract	Window, Forecasting
Models and parameter search	Models	Model Selection
Evaluation and tests	Evaluation	Metrics, Tests
Interpretation, output, and reporting	Interpretation	Output, Reporting

Orientation

Custom Hooks

Package Configuration

Data Pipeline

Feature Pipeline

Forecast Pipeline

Evaluation and Testing

Explanation and Delivery