A monthly sub-national Harmonized Food Insecurity Dataset for comprehensive analysis and predictive modeling
Melissande Machefer, Michele Ronco, Anne-Claire Thomas, Michael Assouline, Melanie Rabier, Christina Corbane, Felix Rembold · 2025
Read A monthly sub-national Harmonized Food Insecurity Dataset for comprehensive analysis and predictive modeling by Melissande Machefer, Michele Ronco, Anne-Claire Thomas, Michael Assouline, Melanie Rabier, Christina Corbane, Felix Rembold (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2501-06076.txt.
Introduction
Melissande Machefer and co-authors build, in a 2025 paper, an open data set that joins the main world food security gauges in one file. Food crises need fast, full data to track and to foretell, yet the gauges sit in split stores with split place codes. Their set, HFID, merges four sources: two phase scales for crisis grade, and two survey scores on what homes eat and how they cope. It is updated each month and uses one shared list of local units, so places line up across sources.
The set spans wide ground and long time, at the level below the nation. Aid groups can use it to watch risk and to plan help. The team also shows where data are thin, so gaps can be closed. Model builders gain a clean base for forecast tools.
Catalog record
| CREATOR | Melissande Machefer, Michele Ronco, Anne-Claire Thomas, Michael Assouline, Melanie Rabier, Christina Corbane, Felix Rembold |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 31 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2501-06076
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
A monthly sub-national Harmonized Food Insecurity Dataset for comprehensive analysis and predictive modeling Melissande Machefer, Michele Ronco, Anne-Claire Thomas, Michael Assouline, Melanie Rabier, Christina Corbane, Felix Rembold Abstract Food security is a complex, multidimensional concept challenging to measure comprehensively. Effective anticipation, monitoring, and mitigation of food crises require timely and comprehensive global data. This paper introduces the Harmonized Food Insecurity Dataset (HFID), an open-source resource consolidating four key data sources: the Integrated Food Security Phase Classification (IPC)/Cadre Harmonisé (CH) phases, the Famine Early Warning Systems Network (FEWS NET) IPC-compatible phases, and the World Food Program’s (WFP) Food Consumption Score (FCS) and reduced Coping Strategy Index (rCSI). Updated monthly and using a common reference system for administrative units, the HFID offers extensive spatial and temporal coverage. It serves as a vital tool for food security experts and humanitarian agencies, providing a unified resource for analyzing food security conditions and highlighting global data disparities. The scientific community can also leverage the HFID to develop data-driven predictive models, enhancing the capacity to forecast and prevent future food crises. Background & Summary Achieving the United Nations Sustainable Development Goals 2 – preventing food shortages and reducing global hunger – is an imperative endeavor [ 1 ] . A critical step toward this ambitious goal is the establishment of continuous and reliable monitoring of food security situation, striving for the broadest geographic coverage and the highest temporal frequency possible. Anticipating and understanding the evolution of food security is essential for effective policy-making. Despite the need for consistently updated data, especially in remote and often most vulnerable areas, this objective remains challenging. Significant efforts have been made by various institutions and networks towards providing the essential data needed to inform strategies that can preempt and alleviate food crises and with tools to visualise [ 2 ] and monitor the situation. Notable among these are the campaigns and methods supported by the Cadre Harmonisé (CH), the Food and Agriculture Organization (FAO), the Famine Early Warning Systems Network (FEWS NET), the Integrated Food Security Phase Classification (IPC), the World Bank (WB), and the World Food Programme (WFP). However, monitoring food insecurity at a global level faces significant challenges due to the uneven coverage of countries and regions. Moreover, disparate criteria for measuring food insecurity are employed, depending on the goals of the stakeholders. The complexity of food insecurity, influenced by context specific factors, challenges the development of a universal, objective measure, due to its inherent diversity and multi-dimensionality [ 3 ] . The frequency of data collection and the periods monitored of the various food security indicators often vary, reflecting the differing needs of each initiative at local level. Additionally, geographical inconsistencies arise from the use of various reference systems for administrative boundaries and locations. Consequently, the current landscape of food insecurity assessment is marked by a lack of cohesion, highlighting the need for a more integrated and systematic methodology at a global level [ 4 , 5 , 6 , 7 , 8 ] . Data scientist practitioners have attempted to leverage publicly available datasets [ 2 ] , proposing models for nowcasting and forecasting a stand alone target indicator [ 9 , 10 , 11 , 12 , 13 , 14 , 15 , 16 , 17 , 18 , 19 ] . Using multiple proxies and targets to assess food security would be more effective than a single measure, as it yields a fuller, multidimensional understanding of food insecurity. However, some food insecurity datasets remain unused, due to recent availability, but also the complexity of the definition and of the technical direct access. To bridge these gaps, we have developed the Harmonized Food Insecurity Dataset (HFID), a comprehensive compilation of major available sub-national food insecurity data sources with updates on a sub-annual basis. Our dataset spans from 2007 to the present and encompasses four key variables: the IPC/CH phases [ 20 ] , the FEWS NET-IPC compatible phases [ 21 ] and prevalences of population with insufficient food consumption [ 2 ] based on single indicators of food consumption: the Food Consumption Score (FCS) and reduced Coping Strategy Index (rCSI), both sourced from WFP [ 22 ] through two different methods. Records within the dataset correspond to the first or second administrative level of the countries, harmonized using the Global Administrative Areas (GADM) database as the standard reference for the nomenclature of administrative units [ 23 ] . The HFID comprises 311,838 records, encompassing 5,508 units and 1,264 units respectively corresponding to the second and first administrative division across 80 countries. By integrating these variables within a unified temporal and spatial framework, we enable a thorough and unparalleled comparative analysis of food insecurity trends derived from various datasets. The HFID not only provides the ability to examine historical data sourced from different providers but also stands as a powerful instrument for ongoing monitoring of food insecurity using both assessments from multi-partner analytical approaches (e.g. IPC/CH) and single outcome indicators (e.g. FCS). It presents a centralized hub for accessing a range of crucial variables, thus streamlining data management and improving the precision and impact of data interpretation. Additionally, it serves as a vital tool for identifying data availability and critically, the areas where data is lacking, encouraging more targeted data collection efforts in underrepresented areas or countries. Finally, the HFID stands as a valuable resource in its own right while also offering significant potential to propel forward the development of advanced modeling techniques within the field of food security. Most studies have targeted the prediction of single indicators. There seems to be an absence of efforts to create forecasts that combine various food insecurity indices, utilizing either multi-output models or the construction of an aggregate target index [ 24 ] . It would therefore be intriguing to enrich the HFID’s target variables with predictors that encompass the recognized drivers of food crises, in order to associate different food insecurity targets to the varying impacts of these predictors [ 25 ] . This could indicate a multitude of underlying mechanisms at play and of food insecurity dimensions. Exploring these applications, we expect a considerable improvement in our understanding of food insecurity and the complex interplay between elements like weather, climate, conflict, and other disruptive events. Methods Several food insecurity raw data sources have been used to build the HFID. We collected the IPC/CH and FEWS NET-IPC compatible current situation phases through the freely accessible APIs [ 26 , 27 ] , respectively referred hereafter with phase-IPC/CH and phase-FEWS. WFP prevalence of population with insufficient food consumption based on single indicators FCS and rCSI, were instead retrieved from two different sources: a publicly available historical dataset [ 28 ] ) from the Literature (LIT) covering 2006-2021, called in the following WFP-LIT, and a dedicated data dump with all historical records from Real Time (RT) monitoring also available (only last 500 days) in the HungerMap [ 29 ] , called in the following WFP-RT. We hereafter refer to the population prevalence of insufficient FCS (poor and borderline) and crisis and above rCSI ( \geq 19 ) 1 1 1 These thresholds used for computing population prevalence of insufficient food consumption for FCS and rCSI indicators are the ones used