Geolocation and Mapping Learnings
While working for a large government client we were keen to get them to maximise the benefits of their new SAS Viya installation. SAS Visual Analytics is a part of SAS Viya, which, among other things, was being used to present data and results in compelling ways for a variety of stakeholders. The platform, with its advanced data visualisation techniques, enables complex functions to be achieved via a simple drag and drop web interface that is further aided by autocharting. A section of those data visualisation techniques are related to geographical context, where traditional data sources (transactional, customer, operational, etc.) are joined with location data. To be able to make use of the geographical features we needed to define the users requirements, obtain and configure the system, highlight the functionality through demonstration, and train users in how to use and maintain the system.
When given an open remit it is easy to jump in and do what you think the customer wants. This approach can be fantastic at first if there are some quick deliverables and the customer does not have initial requirements due to not knowing the ‘art of the possible’. But as the work continues this unplanned approach can quickly lead to an unending list of deliverables where it is hard to define the work as being complete. It can also result in lost effort implementing a solution that is not required by the customer. Established frameworks like Agile really help to structure the development and critically aid the communication between the development and the customer.
In our recent governmental client work, through interviews with key users we ascertained that the user did not simply want to view data elements on a map but they wanted to be able to group the data in definable regions/boundaries. The regions they were initially interested in were Parliamentary Constituencies and postcodes. Further clarification determined that required postcodes were actually the postcode areas (124 boundaries) and postcode districts (2,982 boundaries) and not the sectors (11,204 boundaries) or full postcodes (1,765,422 boundaries) that were originally requested, knowing this helped with scaling and performance tuning.
The IT system needed to be considered holistically, with training, support and maintenance often not being fully considered. Particularly with mapping, the reference data is regularly updated and processes need to be implemented to ensure the systems are kept up to date.
While this work was conducted on SAS Viya, the basis of a holistic view to system integration/configuration and clear requirements are applicable to many deployments.