At Knoema, all datasets are regularly updated to show the latest data released by the original data sources.
In some cases, datasets are updated by simply adding new data to the existing dataset while in other cases - by uploading a new dataset version. The second case takes place when there are changes in the dataset structure or methodology.
To see the historical versions of a dataset, open a dataset in the Dataset Viewer, and switch to the Archive tab in the right panel. Here, you can see the list of all historical dataset versions and can access them by selecting any link.
Historical dataset versions may be useful when, for example, you want to check the accuracy of forecasts, analyze the history of revisions, or get data for an older time period:
Outlooks. Datasets that contain forecasts are particularly useful for publishing with revisions. For example, the IMF World Economic Outlook is published bi-annually and we have all releases in Knoema dating back to April 2010. Users are notified when they open an older version through their favorites, data search, a dashboard, or some other means and can click to instantly open the newer edition.
Major structural, content changes. Sources sometimes publish similar datasets but with discontinuation of certain data that they no longer collect and/or publish, and we, therefore, retain the prior version and create a new edition so that users can still benefit from the historical data. The Bank of Mexico publishes a useful example of this case titled, Mexico: Resources and Obligations of other Financial Companies-Warehouses of Deposit, Monthly Update.
Alternate base periods. Sources may also release separate datasets covering distinct base periods. In these instances, we will also create separate datasets. One such example is China: Price Index - Producer Price Indices for Industrial Products by Sector published by China’s National Bureau of Statistics. One dataset covers the period 2000 to 2011 and a second covers 2012 to present.
Custom requests. Our clients also often maintain custom versions of datasets in their Enterprise Subscriptions, potentially to remove certain topics, add new measures, or other adjustments that make it more sustainable to maintain separate versions and avoid altering the original source data.