Welcome to macroforecast#
A pandas-first framework for reproducible macroeconomic forecasting workflows.
macroforecast turns a forecasting study into one declarative specification and runs it as a leak-aware pseudo-out-of-sample (POOS) experiment.
A broad set of data transformations. Many ways to turn raw macro series into model inputs.
A wide range of forecasting models. From simple benchmarks to modern machine learning.
Many evaluation tests and interpretation tools. Compare forecasts and understand what drives them.
Reproducible paper replications. Rebuild published studies from start to finish.
Minimal end-to-end pipeline in a few lines.
Concept pages for every workflow stage.
Feature steps and every registered model.
FRED-MD, FRED-QD, FRED-SD, and combined loaders.
Published paper replication studies.
CSSED, fluctuation, PIT, and forecast-path exhibits.
Plug in your own datasets, features, models, tests, and outputs.
Workflow reference grouped by responsibility.
Definitions of every core abstraction.
Citation metadata for papers, reports, and replication packages.