Signal Extraction and Forecasting

A collection of novel forecast approaches: MDFA, M-SSA, Look Ahead DFP and PCS. Explainable AI: xAI.

Click on any of the coloured buttons to access to github repositories (tutorials).

OPEN SOURCE RESEARCH ON FORECASTING

MDFA: multivariate direct filter approach implements Accuracy-Timeliness-Smoothness (A-T-S) forecast trilemma in frequency-domain; M-SSA implements A-S dilemma in time-domain; Look Ahead DFP and PCS implement A-T dilemma in time-domain. All approaches emphasize explainability (closed-form, unique and fast solutions, simple novel principles). xAI: novel explainability tool for Neural Nets applied to time series.

Papers/Books

ArXiv 

R-Code

Github

Tutorials

Github.