🔬 方法论

Data Collection

Raw demographic data collected from WHO Global Health Observatory, PAHO Health Information Platform, and national statistics institutes (INE/IBGE/DANE/etc.). Harmonized to annual frequency, 2000-2026.

Data Cleaning

Outlier detection, missing value imputation (LOCF + linear interpolation), unit standardization (per 100,000 inhabitants), and country ISO code normalization.

Indicator Computation

Core NCRA indicators computed: aging_index = (pop_65+/pop_0-14)×100, dependency_ratio = (pop_0-14+pop_65+)/pop_15-64×100, demographic_dividend_index = standardized composite.

NCRA Risk Score

Composite score (0-100) integrating normalized indicators with empirically-derived weights: aging (30%), dependency (25%), DDI (25%), population structure (20%).

Validation

Cross-validation against WHO reference values, internal consistency checks (Cronbach α > 0.85), and expert review (F11 Gate).

Publication

Datasets published on Zenodo (DOI), GitHub (code), Kaggle, Hugging Face. All formats: CSV, Excel, SPSS, Stata, R, JSON. CC0 1.0.

Aging Index

Range: 0-500+. Values >100 indicate more elderly than children.

Dependency Ratio

Range: 0-200. Higher = more dependents per working-age person.

NCRA Risk Score

Range: 0-100. Higher = greater neuroeconomic cognitive risk.