Jumpstarting the Use of Predictive Analytics
My last post talked about the factors dragging the adoption of Predictive Analytics. This one is about jumpstarting it.
Organisations have too many fears with regard to predictive analytics - can they handle it, can they afford it, will it work... etc. They need to dump this baggage. In truth, there are several ways to embrace predictive analytics, some simpler than expected. Rather than get bogged down by the enormity of the task, organisations must take a level-headed look at their processes and available data - which are the keys to determining the actual solution.
As mentioned in my previous blog post, statistical analytics experts are hard to come by. But they're not the only ones for the job. Research shows that companies have successfully deployed a combination of IT developers, business analysts and users into their BI projects. Only one in two hire specialist analytical modellers. This shows that the unavailability of highly skilled resources need not be a stumbling block. In fact, multi-skilled users bring a distinct advantage of balance, because they will likely recommend a mix of processes and tools that are suited both to regular as well as 'power' users.
The success of predictive analytics hinges on the quality and availability of data. Hence, it is important to assess data availability and associated risk at the outset. Here again, the use of a multi-disciplinary team can improve the accuracy of data validation and eliminate the bias that usually creeps in when only a single analyst is entrusted with the job. Through these measures, the organisation can streamline its sources and quality of information well in advance, rather than grapple with issues midstream.
Having realistic expectations is equally important. Analytics is a long term game, which can yield handsome rewards to those investing in it. Organisations must therefore look at the big picture rather than quick wins to avoid being sidetracked. A vital part of this exercise is correctly defining the business problem which predictive analytics is meant to solve, and making sure that the latter is indeed the best fit for this situation.
The good news is that vendors are doing their bit to ease adoption of analytics solutions. They have developed plug and play applications with rich, customised functionality that are an expedient and cost effective alternative to building an in-house solution from scratch. Many vendors offer open specialised solutions which can be built on to existing packaged products as a differentiating element. What's more, many of these solutions are hosted on the cloud, and made available to users on a 'pay as you go' basis, thereby reducing their initial investment significantly, as well as divesting them of the burden of maintaining the solutions in-house.
