Predictive analytics is all over the news nowadays. It is touted as the new solution to all of the problems facing the business world! Suddenly, anybody who is not talking about it feels they are being left far behind in the race. And hence a spurt of activities around analytics (Gartner Study: Hype Cycle for Analytic Applications, 2011 and http://www.informationweek.com/news/software/bi/231903316“)
What is Predictive analytics? Predictive analytics is used to determine the probable future outcome of an event or the likelihood of a situation occurring. It is the branch of data mining concerned with the prediction of future probabilities and trends. Predictive analytics analyzes large amounts of data with different variables; it includes clustering, decision trees, market basket analysis, regression modeling, neural nets, genetic algorithms, text mining, hypothesis testing, decision analytics, and more.
All of the above while being highly technical and data intensive, help empower decision makers at all levels in the organizations
- Operational decisions: Should the Banker extend credit to the customer, if yes how much credit? Some predictive models are automated and integrated to operational systems, hence if you have ever had a credit card blocked when you were traveling, it is because your unusual transactions have triggered automated rules set by anti-fraud/ anti-identity theft data mining models.
- Strategic Decisions: How much to price a product? What profile of customers to approach to get the best response to the product/offer?
- Executive Decisions: What are the geographies/products that the organization wants to operate in? What is the optimum capital structure? What are the investments the organization should make to keep pace with its forecasted growth?
All of the above decisions are made at all organizations. In the absence of predictive analytics, most of the decisions are made based on user intuition, their business experience and some analysis. What predictive analytics does is provide a data driven way of approaching the above decisions, leveraging the collective wisdom in its data , ensuring uniformity in operational decisions, and helping organizations effectively plan their strategies . Also, old BI, manual analysis and experience will not be sufficient for businesses that operate in scale (geo spread, number of customers & channels) and rapidly changing environment. Business users will need data-driven insights to boost good business intuition
Faster response times to enable decision makers and the increasing need to tailor responses at a granular level (for customers and events) are making PA increasingly complex from a deployment and IT integration perspective. To achieve this level of accuracy in predictive modeling, modelers are leveraging newer sources of data (Big data), and looking to newer and newer techniques or a combination of techniques to gain that competitive edge.
Given these shifts, IT departments are going to have to revisit how they support the predictive modeler and the predictive modeling environment, A robust data infrastructure that accommodates the predictive modeler’s needs will reduce modeling time and an effective integration to decision making system will be the key to harnessing the power of predictive analytics to gain a competitive edge.
…In a lighter vein, without my thinking Cap on!
What would you predict would be the minimum number of people needed in a random group for a 100% chance of one common birth date? Or 98% chance of one common birth date?
If you said 366, welcome to the world of statistics, and look up the solution on Wikipedia (Birthday problem). It is a popular problem and the answer continues to amaze people