Text mining and machine learning reveal global determinants of food insecurity
Authors:
Open Access
- 21 October 2025
- journal article
- research article
- Published by Springer Nature in Scientific Reports
- Vol. 15 (1) , 1-19
Abstract
This study applies Natural Language Processing (NLP) and Machine Learning (ML) to investigate global historical trends in food security. Using USAID’s Famine Early Warning Systems Network’s (FEWS NET) comprehensive reports spanning over two dozen countries, it explores prevalent dimensions such as shocks, outcomes, and coping capacities, offering insights into long-term food security conditions. Results highlight the prevalence of market and climate impacts across the countries and period considered. Based on results from the topic classification, ML models were applied to determine the most important factors that predict food insecurity. The analysis confirmed market shocks as the main predictors of food insecurity globally. The approach demonstrates the potential for extracting valuable insights from narrative sources that can support decision-making and strategic planning. This integrated approach not only enhances understanding of food security but also presents a versatile tool applicable beyond the context of humanitarian aid.Keywords
Funding Information
- United States Agency for International Development
- CGIAR Initiative on Climate Resilience, ClimBeR
- CGIAR Initiative on Fragility, Conflict and Migration
This publication has 53 references indexed in Scilit:
- Fundamentals of Supervised Machine Learning2023
- Spatial dynamics across food systems transformation in IFAD investments: a machine learning approachFood Security, 2021
- Big Data and Social Indicators: Actual Trends and New PerspectivesSocial Indicators Research, 2016
- A framework for a regional integrated food security early warning system: a case study of the Dongting Lake area in ChinaAgriculture and Human Values, 2014
- An Introduction to Statistical Learning2013
- Early warning systems for food security in West Africa: evolution, achievements and challengesAtmospheric Science Letters, 2011
- The Elements of Statistical Learning2009
- Modern Applied Statistics with S2002
- Naive (Bayes) at forty: The independence assumption in information retrieval1998
- A translation approach to portable ontology specificationsKnowledge Acquisition, 1993