An executive-grade forecasting system for Ethiopia’s digital financial transformation. Leveraging Interrupted Time Series and causal modeling to predict Findex 'Access' and 'Usage' metrics through 2027, quantifying the impact of Telebirr, M-Pesa, and NBE policy shifts.
# EthioPulse-Forecaster
**Financial Inclusion Forecasting System for Ethiopia**
Analytical system designed to forecast Ethiopia's digital financial transformation trajectory, aligned with the World Bank Global Findex framework. This system supports decision-making for the National Bank of Ethiopia, mobile money operators, and development finance institutions.
## Table of Contents
- Project Overview
- Repository Structure
- Key Features
- Installation
- Usage
- Task 1: Data Exploration & Enrichment
- Task 2: Exploratory Data Analysis
- Task 1: Detailed Documentation
- Description
- Implementation
- Files Created
- Execution Results
- Key Insights
- Task 2: Detailed Documentation
- Description
- Implementation
- Files Created
- Execution Results
- Key Insights
- Data Sources
- Methodology
- Contributing
- License
- Contact
## Project Overview
This repository implements a comprehensive forecasting system focusing on two core pillars of financial inclusion:
- **Access**: Account Ownership Rate
- **Usage**: Digital Payment Adoption Rate
The system analyzes historical trends (2011-2024), identifies structural constraints and leading indicators, and prepares the foundation for time-series forecasting (2025-2027).
## Repository Structure
```
EthioPulse-Forecaster/
├── .github/workflows/ # CI/CD workflows
├── data/
│ ├── raw/ # Original datasets
│ └── processed/ # Enriched and cleaned data
├── notebooks/ # Analysis notebooks
├── src/ # Python modules
├── dashboard/ # Interactive dashboard
├── models/ # Trained forecasting models
├── reports/ # Analysis reports and insights
└── tests/ # Unit tests
```
## Key Features
- **Unified Schema Dataset**: Works with `ethiopia_fi_unified_data.xlsx` containing observations, events, impact links, and targets
- **Event-Driven Analysis**: Pillar-agnostic events with causal logic through impact_link records
- **Data Enri …