Data science project predicting Human Development Index (HDI) in Africa using alternative socio-health indicators and optimizing international aid allocation via linear regression, cross-validation, and constrained convex optimization to maximize quality of life under real-world budget and demand constraints.
# Improving Quality of Life: Optimal Financial Aid Allocation
## Overview
This project applies data science, predictive modeling, and mathematical optimization to design an evidence-based international aid allocation strategy for Africa. The goal is to predict Human Development Index (HDI) using features that can be directly improved through financial aid, and then optimally allocate a fixed budget across resources to maximize quality of life.
The workflow includes:
- Data integration: Merging multi-source global development datasets (health, sanitation, water access, HDI, country mapping).
- Data preprocessing: Country-level cleaning, missing value imputation, normalization, and consistency harmonization.
- Feature engineering: Distribution correction using transformations and engineered feature evaluation.
- Dataset filtering: Continent-level analysis and Africa-specific modeling.
- Predictive modeling: HDI regression modeling using k-fold cross-validation.
- Model evaluation: Performance analysis using train/test R², prediction stability, and coefficient interpretation.
- Optimization modeling: Budget-constrained convex optimization for aid allocation.
- Sensitivity analysis: Binding and non-binding constraint evaluation.
---
## Key Features
### Data Preprocessing
- Multi-source dataset integration (WHO, Our World in Data, Oxford Martin School)
- Country name normalization and consistency handling
- Modular missing-data imputation functions
- Country-level and continent-level filtering
- Structured feature selection
- Removal of non-actionable policy variables
### Exploratory Data Analysis (EDA)
- Continent-level development comparison
- Developing country distribution analysis
- Life expectancy and development indicators
- Africa-focused scope selection
- Distribution analysis and skew detection
- Correlation analysis
### Feature Engineering
- Distribution correction using square-root transformations
- Feature normalization
- Automated feature combinatio …