Parametric crop insurance platform for Ghana using satellite rainfall data (CHIRPS), soil properties (iSDA), and HDX hazard indices to predict district-level drought risk with XGBoost. Features an interactive Streamlit dashboard with choropleth mapping and simulated payout triggers. Built on Google Colab.
# 🌾 Parametric Crop Insurance Platform — Improving Crop Insurance Using Big Data in Rural Areas
> A Big Data and Machine Learning project for predicting agricultural losses from
> weather events using satellite-derived indices and historical climate data.
---
## 📌 Project Context
**Course:** Advanced Big Data Management
**Programme:** MSc Internet of Things & Big Data
**Institution:** Ghana Communication Technology University
**Academic Year:** 2025/2026
---
## 🔍 The Problem
Smallholder farmers in rural Ghana — and across Sub-Saharan Africa broadly — are among the most economically vulnerable populations in the world. Their livelihoods depend almost entirely on rain-fed agriculture, yet they operate with virtually no financial safety net against weather shocks.
**Conventional crop insurance has failed this demographic for three structural reasons:**
### 1. Verification Dependency
Traditional indemnity-based insurance requires physical field inspections to verify crop losses before any payout is made. In rural Ghana, where road infrastructure is limited and farm plots are geographically dispersed, this process is prohibitively expensive, slow, and logistically impractical. By the time an assessment is completed, the farming household has already incurred irreversible financial damage.
### 2. Data Scarcity
Actuarial pricing of agricultural insurance requires robust historical yield and loss data. Most smallholder farms in Sub-Saharan Africa lack formal records. Government and institutional yield databases are sparse, inconsistent, or inaccessible — making it impossible for commercial insurers to price risk accurately. This data gap causes insurers to either price premiums too high for farmers to afford, or exit the market entirely.
### 3. Basis Risk
Even where insurance products exist, they often fail to reflect localised weather conditions. A national or regional average rainfall figure may show adequate rainfall for a given season while specific farmin …