melat33/Ethiopia-Financial-Forecast

# Ethiopia Financial Inclusion Forecast **Forecasting Ethiopia's digital financial transformation — built for a consortium of development finance institutions, mobile money operators, and the National Bank of Ethiopia.** **→ Live Dashboard** · **→ API Docs (Swagger)** > T --- ## The problem Ethiopia's digital financial system grew explosively — Telebirr passed 54 million users, M-Pesa entered the market in 2023, and P2P digital transfers recently overtook ATM cash withdrawals for the first time. Yet Global Findex data shows account ownership grew only **+3 percentage points** between 2021 and 2024, a sharp deceleration from the +11pp and +13pp gains of prior rounds. The consortium needed to know: what's actually driving inclusion, and where is it headed through 2027? ## What this project delivers An end-to-end forecasting system — not just charts, a full pipeline with a validated model behind it: - **Enriched dataset**: the starter data had zero `impact_link` records connecting events to outcomes and no observation for the Usage/digital payment forecast target. Both gaps were identified, sourced, and closed — documented line-by-line in `data_enrichment_log.md`. - **A validated causal model, not just a fitted curve**: the event-impact model's estimate for Telebirr + M-Pesa's effect on mobile money adoption (9.14%) lands **0.31 percentage points** from the actual observed 2024 figure (9.45%) — checked against real data, not asserted. - **Honest forecasting**: base-case Access is projected at **54.6% for 2025**, a **15.4pp shortfall** against the NFIS-II policy target of 70%. Where the model rests on unvalidated, forward-looking assumptions (events that launched in 2024-2025 with no post-period data yet), that's flagged explicitly rather than hidden behind a confident-looking number. - **A real product, not a notebook**: FastAPI backend, React dashboard with a custom design system, deployed and live at the links above. ## Key findings | Question …