Predicting the future of predictive BI

There is this mysterious fascination with knowing the future. Children of the 80s all dreamed of owning that Gray's Sports Almanac... the one that Marty McFly acquires in the year 2015 in 'Back to the Future II'.

People dream, strive for, and continuously attempt to accurately tell the future.

Yolanda Smit is senior BI business analyst for PBT.

Today still, in business, people dream, strive for, and continuously attempt to accurately tell the future, and that is where predictive BI is becoming more and more mainstream, especially in the financial services, mining, and retail industries in South Africa.

Predictive analytics is strongly applied in a few key areas and is definitely picking up momentum in others. In sales and marketing, customer churn and response models are becoming pervasive, and I've seen it quite successfully implemented to segment customers, specifically for direct marketing campaigns.

In risk assessment for credit, the interest in predictive BI was spurred with the impact of the recession and the NCA hitting the South African financial services industry at almost the same time. “Sellers of money” had to, very smartly and accurately, predict their risk on an individual loan application level to maximise turnover while containing risk.

Fraud-prevalent industries are using predictive BI to detect and prevent possible fraudulent transactions. These fraud prevention applications are most likely already mainstream in the financial services and insurance industries.

I'd extend the two years to a maximum of five years for mainstream adoption, but there are definitely industries in South Africa that are keeping up with the global leaders, if not already leading globally in some instances.

There are three key trends observed in the predictive BI space that clearly indicate the increased maturity:

There is a higher and wider adoption across the enterprise user base. The demand for predictive BI no longer comes only from the statisticians. An increasing number of data and business analysts are starting to grasp the value of predictive BI and are increasingly demanding this value too. This implies that ease of use has to become a specific goal for all predictive BI tools. Remember that predictive analytics can provide significant top-line benefits to companies, but in the wrong hands it can lead to distracting or misleading results. If a company is planning to purchase predictive analytics tools or solutions, it is important to consider the skill level of the people using the technology and the range.

Predictive models are more pervasively being embedded as part of operational processes.This provides great advantages when the model, along with some decision rules, can be inserted in, for example, the company's claims processing system in order to flag claims with a high probability of fraud. In other cases, the model itself might be totally hidden to the end-user. For instance, a model could be built to predict customers who are good targets for upselling when they call into a call centre. The call centre agent, while on the phone with the customer, would receive a message on specific additional products to sell to this customer. The agent remains totally oblivious to the predictive model that is working behind the scenes to make this recommendation.

Predictive BI will have to be geared for 'big data'. As companies gather greater volumes of disparate kinds of data (ie, both structured and unstructured), they are looking for solutions that can scale. Real-time analysis of large amounts of data is also becoming more prevalent. A specific source of big data is social networks, and this opens up new opportunity for predictive BI, referred to as social analytics. This is the movement whereby social networking behaviour of customers are monitored, for example, to derive their sphere of influence, predict their propensity to a product, or even profile a customer's risk or fraud probability based on his social networking behaviour.

Flowing from the 'big data trends', a whole new frontier of customer or market insight is dawning through predictive BI, called sentiment analysis. This is the process of gathering unstructured data from social networks and doing text mining in order to derive the market's perception towards an organisation, product or service.

Is a company mining for tweets about it? Is it analysing the general sentiments in the market about the company's product leveraging the social networks? Is the company inferring its customers' needs and wants by monitoring and mining its Facebook status?

There was an interesting response when this subject was broached at the last ITWeb BI Summit. It seems there is a resistance to this kind of analysis, motivated by ideas like privacy and confidentiality, to name a few.

So it is with this that I challenge South African businesses - global competitors have already tasted the first fruits of return on investment in sentiment analysis. Are local businesses courageous enough for this new frontier? Are they ready to go back to the future? Social networking is the way of the future, and more and more companies are jumping on this proverbial bandwagon. Get ready to join them!