panasheandrewpanaishe1/Repi-Economic-Index

REPI — Real-Time Economic Power Index. AI-driven economic & energy index for Zimbabwe. POTRAZ AI4I 2026, Track 3: Development. # REPI — Real-Time Economic Power Index **POTRAZ AI for Impact Challenge 2026 — Track 3: Development** Project ID: REPI-01 · Team: StrucLeaf REPI Team ## Problem Zimbabwean policymakers, investors and businesses currently assess economic and energy conditions using fragmented, lagging sources — monthly or quarterly bulletins from RBZ and ZIMSTAT, informal market signals, and sector-specific reports that rarely speak to one another. There is no single, continuously updated reference point that synthesises monetary, energy and infrastructure signals into one number decision-makers can track week to week. **Target users:** government planning units, financial institutions and investors, and StrucLeaf's own downstream AI agents (CAPITAL, RISK, POLICY) and the AFRIVIEW broadcast network. ## Solution REPI is a live, AI-computed composite index that fuses public economic data (RBZ, ZIMSTAT, World Bank, FX/fuel reference feeds) with StrucLeaf's own platform telemetry (AEOS/Crystal Solar energy data, Wazi Infra infrastructure signals, AFRIVIEW network activity) into a single tracked score, updated continuously and accompanied by a plain-language briefing explaining what is driving the current reading. ## Demo - **Live dashboard:** repi-dashboard.netlify.app - Video walkthrough: _to be added_ ## Architecture REPI is a seven-layer pipeline: data ingestion → storage → index computation → narrative synthesis → API layer → application surfaces (public dashboard + downstream StrucLeaf agents). Full diagram: `docs/architecture.png` **Core AI components:** - **Composite Index Model** — gradient-boosted regression for dynamic component weighting, plus seasonal-trend decomposition for drift/anomaly detection. This isn't a fixed-weight formula: the relative influence of each input (energy output, FX signals, infrastructure activity) shifts over time, and a static formula would drift out of calibration as the underlying economic structure changes. - **Forecast Module** — shor …