Big analytics is the design, development, monitoring, management and modification of analytical models, where the volume of models, their variety, and the speed with which they need to be deployed, cannot be accommodated by traditional analytical methods. To be successful, big analytics needs to take place within the context of a Decision Management infrastructure (see Wikipedia definition).
Traditional methods include Business Intelligence where analytics is primarily a manual process, typically looking in the rear view mirror for diagnostic purposes (why sales were low in a particular region last quarter). BI is also used to establish the current status of operations and in some cases to establish trends for extrapolation into the future.
BI platforms tend to be fairly dumb in nature, and the user is required to select items of interest and infer meaning from various reports, graphs and charts. BI has a long way to go before the real needs of businesses can be met, and we await the arrival of intelligent BI platforms that provide much lower latency between need and solution.
Predictive analytics platforms are rapidly embracing intelligence to automatically select meaningful attributes and algorithms for analysis, to automate the predictive model lifecycle, to manage large numbers of predictive models, to automate documentation and provide an overall environment for governance. The models do after all make decisions on behalf of the organisation. Some suppliers are more advanced in this respect than others.
The big picture here is the rapidly accelerating need to automate business decisions. Just as we have spent nearly half a century automating transactions and processes, so there is a very large increment in productivity to be gained from decision automation. To this end some firms (typically in financial services) have already automated many decision (loan approval, fraud detection, targeted marketing etc.) – but even here there is a long way to go. And so there is a need to embrace big analytics technologies so that the cost of analytics models is reduced, and the speed of deployment increased.
Nothing less than a full-blown Decision Management platform will allow this, and just as Enterprise Resource Planning (ERP) solutions consolidated transactional activity, so a Decision Management platform consolidates what would otherwise be fragmented point solutions with unacceptable overheads.
Big data is a cost – big analytics is how we generate value.