Forecasting Ethiopia’s financial inclusion (Access & Digital Payment Usage) using time-series analysis, event impact modeling, and interactive dashboards based on Global Findex and national data.
# 🇪🇹 Ethiopia Financial Inclusion Forecast
**Evidence-based forecasting and interactive analytics for financial inclusion in Ethiopia**
Built with **Python • Pandas • NumPy • Scikit-learn • Streamlit • Plotly**
---
## 🎯 Project Overview & Impact
### The Challenge
Ethiopia is experiencing a critical disconnect between **financial service supply and demand**.
Despite massive mobile money expansion:
- **65+ million mobile money accounts registered since 2021**
- **Only 49% of adults report having a financial account (2024)**
- Growth slowed to **+3pp (2021–2024)** vs **+11pp (2017–2021)**
This **73% deceleration** reveals a fundamental gap:
> **Registered accounts ≠ active users**
Policymakers, regulators, and operators need **evidence-based insights** to understand:
- What actually drives financial inclusion
- Which events matter most
- What the future looks like under different scenarios
---
## 💡 The Solution
This project delivers a **production-grade financial inclusion analytics and forecasting system** that:
- 📊 Explores historical access and usage trends
- 🔍 Quantifies the impact of major events (policies, launches, reforms)
- 🔮 Forecasts financial inclusion outcomes for **2025–2027**
- 🎯 Enables scenario planning for policy and investment decisions
- 📈 Presents results through an **interactive Streamlit dashboard**
---
## ✨ Key Features
### 📈 Actionable Policy Insights
Translate complex data into clear forecasts, tables, and visualizations that answer:
- *What will happen?*
- *What should we do next?*
### 🔍 Quantified Event Impact
Measure how **specific events** (e.g., Telebirr launch, M-Pesa entry, FX reforms) affect:
- Account ownership
- Digital payment usage
Delivered via **event–indicator association matrices** with signed impact magnitudes.
### 🔮 Evidence-Based Forecasting
- Forecasts for **2025–2027**
- **Base, optimistic, pessimistic scenarios**
- Confidence intervals and uncertainty discussion
- Sparse-data aware (only 5 Findex point …