By “dimensional analytics”, we mean analytics that requires an understanding of the dimensions of time and space. This paper describes what it is, why it is important, and how to address it.
In an IT context understanding dimensional analytics is non-trivial. This is because there are different types of data, and different processing requirements that are needed, depending on the approach required to both space and time. Specifically, there are differences – if subtle ones – between time-series and temporal data, and there is also a distinction between location-based and geospatial processing. In addition, not only can these different dimensions be combined with each other, they may also be combined with other forms of information such as weather or credit scores or social data to produce, for example, demographic data. Nor is it simply a question of analytics. The way that different types of dimensional data are stored and/or processed is a key issue, whether in a database or via a streaming platform.
You need specific capabilities for a database to support geospatial, time-series and/or temporal data. And going beyond data storage, there are also developers to consider. While a lot of dimensional analytics will be deployed via a visualisation tool or at a self-service level, there are also applications that may need to be developed to make use of these dimensions and this may require specific capabilities to make life easier for developers. Finally, especially in the case of location-based or geospatial information where do you get this dimensional information from? Apart from things like addresses, it’s not as if it is something that companies will typically generate for themselves, so you will need a third-party provider.
All of these considerations are discussed in this paper.