A Python implementation of the Dynamic Factor Model (DFM) for macroeconomic nowcasting, extending the FRBNY framework (Qian & Bok) with a modern API: real-time vintage management, Kalman-based news decomposition, optional Numba acceleration, caching, and interactive Plotly visualizations.
Nowcasting Indian GDP with a dynamic factor model: estimating current-quarter output from mixed-frequency indicators with ragged-edge data. Python.
End-to-End PyTorch implementation of Chang & Kim's (2026) iVDFM pipeline, Includes: amortized encoder, RegimeNet simplex embeddings, Laplace PriorNet, diagonal AR(p) companion-form dynamics, and injective MLP decoder. Trained via ELBO with reparameterized Laplace innovations. Benchmarked against iTransformer and TimeMixer baselines.
Tests convergence in macro-financial panels combining DFMs and OU processes. Implements static/approximate DFMs, Johansen cointegration, and OU half-life summaries. Features robust HC/HAC inference, VAR stability checks, rolling R2, and PLS preselection. Prioritizes reproducible, publication-ready output.
Nowcasting UK GDP before its official release using monthly indicators (Index of Services, Index of Production) via bridge equation and dynamic factor model
Add a description, image, and links to the dynamic-factor-model topic page so that developers can more easily learn about it.
To associate your repository with the dynamic-factor-model topic, visit your repo's landing page and select "manage topics."