reviser is an R package for analyzing real-time data revisions in macroeconomic time series. It provides a unified workflow for reshaping vintage datasets, computing revision statistics (bias, noise-to-signal ratios, serial correlation tests), identifying the first statistically efficient release, and nowcasting future revisions using state-space models (Kishor-Koenig and Jacobs-Van Norden). The package was peer-reviewed through rOpenSci and ships with GDP vintage data for the US and Euro Area as worked examples.
Table of contents
Why revision analysis deserves its own workflowA compact example with GDP vintagesFrom descriptive analysis to revision nowcastingWhat reviser addsTry it and push it furtherSort: