selamasnake/ethiopia-fi-forecast

# Forecasting Financial Inclusion in Ethiopia ## Project Overview This project builds a forecasting system to track Ethiopia's digital financial transformation using time series methods. The focus is on two core dimensions of financial inclusion, defined by the World Bank's Global Findex: 1. **Access** — Account ownership rate 2. **Usage** — Digital payment adoption rate ### Business Context Ethiopia is rapidly digitizing financial services: Telebirr has over 54M users (since 2021), M-Pesa entered in 2023, and P2P digital transfers now surpass ATM withdrawals. Yet only 49% of adults have a financial account (2024 Findex). Selam Analytics is tasked with: * Understanding drivers of financial inclusion * Assessing impacts of events like product launches, policies, and infrastructure investments * Forecasting Access and Usage trends for 2025–2027 ## Project Structure ### Notebooks * **`schema_exploration.ipynb`** — Explores dataset schema, pillars, record types, confidence, and sources * **`data_enrichment.ipynb`** — Adds new observations, events, and impact links; saves enriched datasets and documents changes in `logs/data_enrichment_log.md` * **`eda.ipynb`** — Performs exploratory data analysis, visualizes trends, correlations, and events; documents key insights, data gaps, and hypotheses * **`event_impact_modeling.ipynb`** — Models how events (policies, product launches, infrastructure) affect financial inclusion indicators * **`forecasting.ipynb`** — Builds trend and event-augmented forecasts for Access and Usage; generates scenario-based projections (Baseline, Optimistic, Pessimistic) ### Source Code (`src/`) * **`eda.py`** — EDA class for loading, preprocessing, summarizing, and filtering data; provides helper methods for Task 2 analyses * **`forecasting.py`** — `Forecaster` class: fits trend models, merges events and impacts, computes cumulative effects, generates scenario-based forecasts * **`impact_model.py`** — `ImpactModel` class: maps impact magn …