Data API

Country Forecasts

6-month and 12-month ahead forecasts, updated monthly on the 1st.

GET https://forecastlab.org/data/forecasts/latest/forecasts_h6.csv
GET https://forecastlab.org/data/forecasts/latest/forecasts_h12.csv
GET https://forecastlab.org/data/forecasts/latest/Hist.csv
GET https://forecastlab.org/data/forecasts/latest/metadata.json

Grid Forecasts (PRIO-GRID)

Spatial forecasts by month. Replace {period} with YYYY-MM (e.g., 2026-03). For bulk downloads, see Grid Downloads below.

GET https://forecastlab.org/api/v1/grid/{period}/points-m1.json
GET https://forecastlab.org/api/v1/grid/{period}/points-m2.json
GET https://forecastlab.org/api/v1/grid/{period}/points-m3.json
GET https://forecastlab.org/api/v1/grid/{period}/points-m4.json
GET https://forecastlab.org/api/v1/grid/{period}/points-m5.json
GET https://forecastlab.org/api/v1/grid/{period}/points-m6.json

Historical Archive

Access any historical forecast period. Replace YYYY-MM with the desired period.

GET https://forecastlab.org/data/forecasts/archive/YYYY-MM/forecasts_h6.csv
GET https://forecastlab.org/data/forecasts/archive/YYYY-MM/forecasts_h12.csv
GET https://forecastlab.org/data/forecasts/archive/YYYY-MM/Hist.csv
GET https://forecastlab.org/data/forecasts/archive/YYYY-MM/metadata.json
GET https://forecastlab.org/data/forecasts/archive/YYYY-MM/forecasts-YYYY-MM.zip

Python

import pandas as pd

# Load latest 6-month forecasts
url = "https://forecastlab.org/data/forecasts/latest/forecasts_h6.csv"
df = pd.read_csv(url, index_col=0)

# Load metadata to understand forecast periods
metadata_url = "https://forecastlab.org/data/forecasts/latest/metadata.json"
metadata = pd.read_json(metadata_url, typ='series')
print(f"Forecast start: {metadata['forecast_start_date']}")

# Get forecasts for a specific country
afghanistan_forecast = df["Afghanistan"]
print(afghanistan_forecast)

R

library(readr)
library(jsonlite)

# Load latest 12-month forecasts
url <- "https://forecastlab.org/data/forecasts/latest/forecasts_h12.csv"
forecasts <- read_csv(url)

# Load metadata
metadata_url <- "https://forecastlab.org/data/forecasts/latest/metadata.json"
metadata <- fromJSON(metadata_url)
cat("Forecast start:", metadata$forecast_start_date, "\n")

# Extract specific country
syria_forecast <- forecasts$Syria

curl / wget

# Download latest forecasts
curl -O https://forecastlab.org/data/forecasts/latest/forecasts_h6.csv

# Download complete bundle for specific period
curl -O https://forecastlab.org/data/forecasts/archive/2026-03/forecasts-2026-03.zip

# Download grid forecast for month 1
curl "https://forecastlab.org/api/v1/grid/2026-03/points-m1.json" > grid_m1.json

JavaScript / Node.js

// Fetch grid forecast data
const response = await fetch(
  'https://forecastlab.org/api/v1/grid/2026-03/points-m1.json'
);
const gridData = await response.json();

// Fetch metadata
const metaResponse = await fetch(
  'https://forecastlab.org/data/forecasts/latest/metadata.json'
);
const metadata = await metaResponse.json();
console.log('Forecast period:', metadata.forecast_start_date);

Metadata (JSON)

Contains forecast period information and data coverage details.

{
  "run_date": "2026-03-01T01:15:43.123456",
  "data_end_date": "2026-02",
  "forecast_start_date": "2026-03",
  "h6_end_date": "2026-08",
  "h12_end_date": "2027-02",
  "training_window_months": 24,
  "historical_start_date": "1989-01",
  "total_historical_months": 445
}

Forecast CSVs

Country-level forecasts with rows as forecast months and columns as countries.

• Rows: 0-5 (h6) or 0-11 (h12), starting from forecast_start_date
• Columns: Country names
• Values: Predicted monthly fatalities
• Variants: mean (default), min, max

Historical CSV

Observed fatality data from UCDP, 1989-present.

• First column: Date (YYYY-MM-DD format)
• Other columns: Country names
• Values: Observed monthly fatalities

Grid JSON

GeoJSON FeatureCollection with point geometries for each grid cell.

{
  "type": "FeatureCollection",
  "features": [
    {
      "type": "Feature",
      "geometry": {"type": "Point", "coordinates": [lon, lat]},
      "properties": {"value": 2.34}
    }
  ]
}

License

CC BY-NC 4.0 (Attribution-NonCommercial). For commercial use, contact us.

Rate Limits

Cache responses locally. Data updates monthly on the 1st only. Excessive requests may be rate-limited.

Update Schedule

Forecasts generated on the 24th and 1st at 01:00 UTC. Data available by 03:00 UTC same day.

Citation

Schincariol, T., Frank, H., & Chadefaux, T. (2025). Accounting for variability in conflict dynamics: A pattern-based predictive model. Journal of Peace Research. DOI: 10.1177/00223433251330790

See full citation guidelines.

Support

Questions, issues, or feature requests: