Analytics platforms have a lot to live up to. The scope may be fairly straightforward, but expectations can be high, and there is a wide range of users and customers, all of whom have slightly different needs. This post explores what IT decision makers can expect from an analytics platform.
The definition of a data science platform is simple. It is a software foundation that is engineered to generate insights from your data in any computing environment. Platforms are – or should be – built on a strategy of using analytical insights to drive business actions. They therefore need to support every phase of the analytics life cycle, from data, to discovery, to deployment. The best platforms are environment-neutral, meaning that they will work equally well whether run from the cloud or on-premises, or even distributed across databases and edge systems.
As is so often the case, however, this definition masks a number of practical issues and benefits for a wide range of groups, including IT decision makers.
Requirements of IT decision makers
IT decision makers have a number of requirements for a data science platform. These include:
The platforms must work equally well from the cloud or on-premises, and also on the edge. Hybrid architectures are also very common, which is why a fast and easy rollout is particularly important.
Cost-effectiveness and elasticity
This may sound obvious, but it is still worth saying. Platforms need to be justifiable in budget terms; they must be cost-effective, and they must help the organisation to meet its business objectives. If the platform cannot be justified, it will not be purchased.
Reliability and performance
IT decision makers need to be certain that the platform will be reliable, which in this context means both stable and performant. If the platform – for any reason – does not provide the desired results exactly when they are needed, it will certainly not be used.
Maintainability and automation
IT decision makers need to know that the platform will be easy to maintain and automate. Most organizations expect to have hundreds, if not thousands, of analytical models within a fairly short space of time. A key benefit of platforms is that they make their management significantly easier by providing model inventory and management, together with validation.
Increasing transparency and traceability, and making it easier to carry out monitoring and auditing activities, is a key IT responsibility. A modern analytics platform has to provide those instruments out of the box.
Security and compliance
These are closely linked to governance issues, but also cover data protection, roles and responsibilities, ethics and codes of conduct. These, too, are key IT responsibilities, so it is no surprise that decision makers need to be confident that the platform will support these essentials.
Reusability and portability of business logic
Relies heavily on governance aspects like transparency and traceability and allows business users to search und reuse already existing assets like reports, data tasks or code across departments or environments. Portability also speeds up the go-to-production process significantly and is therefore among the most important IT requirements.
Implementing an analytics platform has shown great promise in growing the strategic value of analytics and in fostering innovation. These are just two findings from new research consisting of in-depth interviews with professionals in 132 organizations and a global online survey. Download your copy here.
Platforms may be either commercial or open source, but whichever is chosen, it is important that the platform is flexible enough to allow applications to be used from a wide range of sources, including other vendors and open source software, as this increases usability and functionality.
Platforms need to allow a wide range of uses, including various types of powerful algorithms, transformations and visualisations but seamlessly integrated and in line with design principles and policies, to make them maximally usable within the organisation. Everyone is talking about self-service interfaces at the moment. This is because the range of IT-tasks is constantly expanding and IT-staff cannot permanently supervise sandbox projects. To avoid unnecessarily burdening the IT department, data science platforms should also offer business users simple self-service-possibilities for data preparation and exploration.
User adoption and accessibility
IT decision makers’ final requirement is that the platform should be adopted by users. Of course this is hard to predict, but the hope is that it will support collaboration and knowledge-sharing, and therefore pay back the initial investment in multiple ways. Accessibility and approachability are however absolute key to ensure enduser-buy-in.
Many of these requirements, of course, are closely linked. For example, the ease of adoption is dependent on a number of other requirements, including functionality, accessibility, openness and reliability, as well as overall performance. Governance, security and compliance are also closely connected. It is, however, worth considering them separately, because doing so will enable them to be assessed more clearly – and ensure that the analytics platform delivers for IT decision makers.
Read more about the SAS® Platform