Méthode et sources

1. Présence opérationnelle (3W : qui fait quoi et où) — The source publishes one row per district, organisation and sector. Organisations are de-duplicated by acronym, name and type; the sectors each one reported are listed together. An organisation counts as national or local when the dataset's organisation type is National NGO or Government. Rows without a district p-code are dropped and counted below.
2. Sécurité alimentaire (IPC / Cadre Harmonisé) — Only rows of type 'current' (an observed analysis, not a projection) at admin level 2 are used. For each district the single most recent reference period is taken and its phase breakdown shown. Rows published at admin level 2 without a p-code are dropped and counted below, never redistributed.
3. Personnes déplacées internes — For each district the rows with the most recent reference period end are taken. The same figure is often published by more than one monitoring operation or reporting round for the same month, so the maximum of the tied rows is used, never the sum, which would double-count.
4. Événements de conflit — Monthly counts for the last 12 published months, kept separately for each event category. ACLED's three categories are NOT mutually exclusive - an event counted as violence against civilians is normally also counted as political violence - so this site never adds them together. The country and district tables use political violence alone as the comparable headline.
5. Financements — Rows without a stated requirement, and rows for reference periods that have not started yet, are dropped; the most recent remaining appeal period is shown. Percentages come from the dataset where published and are otherwise funded divided by required.
6. Besoins humanitaires — Used at country level only, from the Intersectoral rows without a disaggregation category, for the latest reference period. Most countries do not publish these figures at district level.

• Operational presence is self-reported through humanitarian coordination structures. Small, local and informal organisations are the least likely to be listed, so a low count is a prompt to ask locally, not proof of absence.
• IPC / Cadre Harmonisé analyses cover different periods in different countries, and some districts are never analysed. Country totals are the sum of the districts that were analysed, not a national figure.
• Displacement figures come from different monitoring operations with different methods and coverage; they are not comparable between countries.
• Conflict event categories overlap and must not be summed. Event counts reflect reported events, and reporting coverage varies by place and period.
• Gap score uses food insecurity as the only proxy for need, because it is the one severity measure published at district level for all six countries. It ignores health, protection, water and education need.
• District p-codes are the only join key used. Rows without one are dropped rather than guessed.
• Community notes are moderated but unverified. They are labelled as such and kept separate from official data.

Data is obtained from the OCHA Humanitarian Data Exchange (HDX) through the Humanitarian API (HAPI) country datasets. HAPI republishes data produced by the original providers, who should be credited for it.

• Operational presence: national humanitarian coordination structures and OCHA country offices (3W / who does what where).
• Food security: IPC (Integrated Food Security Phase Classification) and, in the Sahel, the Cadre Harmonisé.
• Internal displacement: IOM Displacement Tracking Matrix (DTM) and national partners.
• Conflict events: ACLED (Armed Conflict Location & Event Data Project).
• Humanitarian needs and funding: OCHA (HPC / Financial Tracking Service).
• Baseline population: national statistical offices via OCHA HDX.
• Licences for each dataset are listed in the table above and apply to the underlying data.