ViEWS replication data

Besides the datasets that ViEWS makes available in the download section of the website, in the interest of transparency, all code and data underlying ViEWS forecast is public. In the future, this full data collection as well as all the underlying code will be available interactively both via a web-based download system as well as through a machine-to-machine communication protocol.

However, due to technical reasons, mostly regarding to the size of the database, this cannot be made available on the web right now. Code will similarly be made available on GitHub upon the online publication of the database. If you want access to the full replication data (currently over 200 GB) and code, contact ViEWS at  A description of the database and code follows.

ViEWS stores its data in a large (>200 GB), normalized PostgreSQL database. Full datasets are available directly from ViEWS as signed database dumps (approximately 15 GB in size). Database dumps can be loaded into a PostgreSQL 9.6+ server, and will generate a new database ("views"), replicating the full work environment of ViEWS.

Normalized data is available in the staging schema. A full diagram will be provided with the dump upon request.

Denormalized (flattened) datasets are available in the preflight schema, one for each unit of analysis and temporal resolution, containing all the predictors, without any imputation performed.

  • flight_pgm : priogrid_month
  • flight_pgy : priogrid_year
  • flight_cm : conflict_month
  • flight_cy : conflict_year

Denormalized (flattened) and imputed datasets are available in the launched schema, following similar naming conventions as above, whereas results are published in the landed schema.

Note that data for various components is delivered as flat files (gzipped CSVs) available at the download section.