Danuh16/ethiopia-fi-forecast

# Ethiopia Financial Inclusion Forecasting A forecasting system that tracks Ethiopia's digital financial transformation and projects the two core Global Findex dimensions of financial inclusion: - **Access** — Account Ownership Rate (`ACC_OWNERSHIP`) - **Usage** — Made/Received a Digital Payment (`USG_DIGITAL_PAY`) Built for *Selam Analytics* on behalf of a consortium of development finance institutions, mobile money operators, and the National Bank of Ethiopia. --- ## Status | Task | Description | State | |------|-------------|-------| | **Task 1** | Data exploration & enrichment | ✅ Done | | **Task 3** | Event-impact modelling (`src/impact_model.py`) | ✅ Done | | **Task 4** | Forecasting Access & Usage 2025–2027 (`src/forecast.py`) | ✅ Done | | **Task 5** | Interactive dashboard (`dashboard/app.py`) | ✅ Done | | **Final report** | Blog-post report, PDF (`reports/Final_Report.pdf`) | ✅ Done | ## The unified data schema One table, four record types (see `data/raw/reference_codes.csv`): | `record_type` | `category` | `pillar` | Notes | |---------------|-----------|----------|-------| | `observation` | — | **yes** | a measured value | | `target` | — | **yes** | an official goal | | `event` | **yes** (product_launch, policy, …) | *empty* | pillar-neutral by design | | `impact_link` | — | **yes** | joins an event → an indicator via `parent_id` | **Key principle:** events are *not* pre-assigned to a pillar. Their effects are modelled through `impact_link` records, so the data stays unbiased. ```python from src.data_loader import load_enriched_data, events_affecting_pillar df = load_enriched_data() events_affecting_pillar(df, "ACCESS") # what drives Access, joined to event details ``` ## Datasets | File | Records | Description | |------|---------|-------------| | `data/raw/ethiopia_fi_unified_data.csv` | 57 | Starter (30 obs, 10 events, 14 impact_links, 3 targets) | | `data/raw/reference_codes.csv` | 71 | Valid codes for every categorical field | | `data/pr …