Predictive analytics – everyone’s doing it

No longer the preserve of the IT department or specialist teams, predictive analytics are increasingly being independently deployed in many other parts of organisations

No longer the preserve of the IT department or specialist teams, predictive analytics are increasingly being independently deployed in many other parts of the organisation, and understandably so. Data proliferates and while volume and complexity can slow things down, business demands immediacy. Decision makers want to be ‘doing analytics’ for gain today, not held back by model building for tomorrow.

Marketing has to make sense, on the fly, of ever increasing sources of edge information about customers; Sales needs to work out how best to juggle resources and prioritise channel support; Purchasing and Production must balance lean inventories against business need; and Finance wants cross-organisation insights into profitability and to manage compliance and risk. These operations want to be able to turn data into insights now.

And we are not just talking about heads of operations and other senior executives here. Employees in non-managerial positions are also now using analytical approaches to help them do their jobs smarter and faster. Throughout the organisation, managers and others – including at board level – are becoming acquainted with a variety of analytics and data management packages as a matter of course. Some of these applications will be hosted and served by the enterprise but others will be individually procured from different vendors and could be open source or based in the cloud. To these users, it doesn’t matter which.

Because these individual line-of-business users are not going to be data and analytics experts – and without denying them their initiatives – how does the IT department ensure the data they are using can be trusted, that their chosen tools are suitable and reliable, and the models deployed will deliver actionable results? How does IT ensure these disparate activities talk to each other to contribute collective benefit across the organisation? And not least, what about security?

A one stop analytics ecosystem

The answer is by developing a single, centrally managed ecosystem for all analytics activity, organisation-wide. This combines data for analysis, tracks all analytics activity irrespective of its location, and facilitates the controlled sharing of models and insights across operations. This all needs to happen on a single platform which provides easy access for all users and enables IT to ensure the integrity of projects.

Importantly, this single inclusive platform needs to be flexible enough to adapt to analysts’ evolving needs, rather than require them to conform to a fixed IT setup. It should combine data management capabilities, analytics functionality and model governance. In particular, it must ensure governance for open source applications that don’t include their own inbuilt governance functionality.

Governance in operation

Ensuring governance calls for installing software that connects all parts of the analytics ecosystem. Models are managed by logging everything about their linage and versioning, together with audit trails and source data records. Models are registered in a central inventory and analysts apply business rules to specify the conditions in which models should be exercised. With everything fully documented and tested, IT has a complete understanding of the business context and can monitor inventory and deployment to help ensure good outcomes. For many organisations, this will mean IT teams needing to work and interact differently with the business.

Read more: The Australian Square Kilometre Array Pathfinder finally hits the big-data highway

For the analysts themselves, this governance platform provides a view of all analytics methods, plus a uniform base of portable code that can be incorporated in any activity and scaled as required. And with the platform in place, there is no problem with analysis on the same data from different languages. Algorithms execute the same analytics code on the same models and enjoy the same advantages of multithreading, distributed computing, common data access and security. This is really no different to accessing a webpage – whether on a laptop, smartphone or tablet, the delivery is the same.

Analytical governance in an enterprise-wide analytics ecosystem is the bedrock for good decision making. The one stop approach offers business analysts a consistent and reliable service, whichever vendors’ tools they prefer using. And with governance in place and standards for data integrity and analytics implementation mandated by business need, everyone works against a common approach, irrespective of their job role and the insights they seek.

We are fond of saying that data is our most important asset – so let’s ensure an environment in which any empowered employee is well equipped to ‘do analytics’ for gain, without having to worry about building and managing solutions.

Sandra Hogan is the Director of the Business Analytics Advisory at SAS Australia and New Zealand

Join the newsletter!


Sign up to gain exclusive access to email subscriptions, event invitations, competitions, giveaways, and much more.

Membership is free, and your security and privacy remain protected. View our privacy policy before signing up.

Error: Please check your email address.

More about SAS

Show Comments