news-decomp¶

Purpose¶

news-decomp explains how new data releases change nowcasts. It follows the New York Fed news-decomposition approach and separates a forecast revision into news, re-estimation, and interaction effects.

Features¶

• Decompose nowcast changes across data vintages.
• Distinguish news from model re-estimation and their interaction.
• Analyse contributions by variable, release, forecast horizon, or vintage.
• Produce contribution tables, summary reports, and visualisations.
• Compare revisions across models and nowcasting exercises.

Quick start¶

from news_decomp import NewsData

data = NewsData(decompositions)
data.summary()
data.plot_contributions()

Provide a vintage-aware decomposition table containing forecast revisions and their component contributions. The package validates the data and returns long-format results for further analysis or reporting.

Repository¶

Read the implementation and full API reference in the news-decomp repository.