Research data should be cited for the same reasons as citing to journal articles and other types of publications require citations. Purpose of citing data is to acknowledge the original producer of the data and to help other researchers find the datasets otherwise enable reuse and verification of data, and allow the impact of data to be tracked.
Ultimately, bibliographic links between datasets and papers are a necessary step if the culture of the scientific and research community as a whole is to shift towards data sharing, increasing the rapidity and transparency with which science advances.
(Source: Research Data Management Training Mantra http://datalib.edina.ac.uk/mantra/)
Recommended format for data citation is: Creator (PublicationYear): Title. Publisher. Identifier.
It may also be desirable to include information about two optional properties, version and resource type (as appropriate). If so, the recommended form is: Creator (PublicationYear): Title. Version. Publisher. Resource Type. Identifier.
For citation purposes, DataCite recommends that DOI names are displayed as linkable, permanent URLs. For example:
Read about THOR a 30 month project funded by the European Commission under the Horizon 2020 programme. It will establish seamless integration between articles, data, and researchers across the research lifecycle. This will create a wealth of open resources and foster a sustainable international e-infrastructure. The result will be reduced duplication, economies of scale, richer research services, and opportunities for innovation.