Modeling of a Multidimensional Data-Driven Approach (MDDA) for Optimized ML Model in Poverty Detection

Abstract / Description

This paper introduces the Multidimensional Data-Driven Approach (MDDA), an optimized machine learning framework for poverty detection that integrates multimodal data fusion, fairness-aware learning techniques, and hyperparameter optimization to improve predictive accuracy and scalability. The framework combines diverse socio-economic data sources, including geospatial information, mobile network metadata, financial indicators, and social media analytics, to overcome limitations of traditional household survey methods. The proposed methodology follows five phases: synthetic data generation and preprocessing, feature engineering and selection, machine learning model development, bias mitigation, and performance evaluation. Experimental validation on a synthetic dataset of 100,000 records demonstrates improved classification accuracy, fairness, computational efficiency, and adaptability, providing a scalable solution for evidence-based poverty analysis and policy decision-making.

Keywords

Suggested Citation