Time series completion
Sometimes a time series can be incomplete (i.e. have holes in the periods).
As explained on the features page, one of the design goals of DBnomics is to redistribute data from providers as-is.
In the above series, the year 2002 is missing, and DBnomics will not inject it.
However you can download the time series from a DBnomics client of a programming language (e.g. Python) and complete the time series by yourself, adding missing periods with
Time series alignment
In order to represent many time series on the same table, they must share a common list of periods.
But sometimes, series distributed by providers are incomplete, like this:
A solution would be to complete both series (cf time series completion) in order to insert missing periods using
NA values, but depending on the case, that would insert a huge number of lines.
There is another solution, time series alignment, which consists of unifying the periods of all requested series, and insert
NA values only for those inserted periods, like this:
|PERIOD||Series A||Series B|
The Web API, via its parameter
align_periods, is able to do that:
Note: the difference between both URLs is the usage of the
As described in the data model, datasets can have releases in DBnomics.
The Web API is able to resolve the reserved release code
latest to the actual release code.
For example, if a dataset has 2 releases
WEO:2019-10, declared in this order,
WEO:latest is resolved to
The endpoints of the Web API that accept a dataset code parameter are able to do this resolution using an HTTP temporary redirection (code 302).
curl -I "https://api.db.nomics.world/v22/datasets/IMF/WEO:latest" HTTP/1.0 302 FOUND Location: https://api.db.nomics.world/v22/datasets/IMF/WEO:2019-10 [...]