Logo Lanfrica

Maki4444444/ethiopia-fi-forecast

Domaine:

socioeconomic

Type de record:

project
Créateur:
Mak
Hôte:
# Ethiopia Financial Inclusion Forecasting Data exploration, enrichment, and forecasting project analyzing financial inclusion trends in Ethiopia (Access, Usage, Affordability, Gender, Quality, Trust, Depth), using a unified event/observation/impact_link schema. ## Project Description This project builds a forecasting pipeline for two of the seven financial-inclusion pillars, **Access** and **Usage** by: 1. Exploring and enriching a starter dataset of survey observations, cataloged events (policy changes, product launches, market entries), and modeled event→indicator impact_links 2. Running exploratory analysis to surface patterns, gaps, and testable hypotheses 3. (Later tasks) Modeling event impacts and producing Access/Usage forecasts with an interactive dashboard ## Setup Instructions ```bash # clone and enter the repo git clone cd ethiopia-fi-forecast # create and activate a virtual environment python -m venv .venv source .venv/bin/activate # Linux/macOS/WSL # .\.venv\Scripts\Activate.ps1 # Windows PowerShell # install dependencies pip install -r requirements.txt # launch notebooks jupyter notebook notebooks/ ``` ## Data Sources | Source | Used for | |---|---| | World Bank Global Findex (2011, 2014, 2017, 2021, 2025 waves) | Account ownership, savings, borrowing, gender splits | | National Bank of Ethiopia (NBE) — Financial Stability Reports, draft NDPS 2.0 | Fraud losses, cyberattacks, national account activity | | IMF Financial Access Survey (FAS) | Access indicators | | ITU DataHub | Mobile data affordability | | GSMA | Mobile money penetration | | UNCDF Ethiopia | Agent network activity rates | | Operator/press reporting (Shega, Birr Metrics, The Reporter Ethiopia, etc.) | Agent counts, product launches, incidents | Every individual data point's exact source, quoted text, and confidence level is documented in `data_enrichment_log.md`. ## The Unified Schema All data lives in one table structure with a `record_type` column that determine …