What is Predictive Analytics?

One of the top topics in management discussions nowadays is utilization of predictive analytics for better decision making. Predictive analytics can simply be defined as the process of studying and learning from data and discovering all sorts of visible and hidden patterns and relationships within data and applying that knowledge to predict future events.

Predictive analytics has been around for years and has been continuously improving in terms of the algorithms used, the efficiency of the code and/or the design of software itself. Some large companies and some government departments have been taking advantage of this technology for years.

So, you say, what’s new? The new thing about predictive analytics is the realization of the following points:

• Low computer hardware costs. Analytics processing is a big computer resource guzzler.

• Availability and collection of huge amounts of data for and about everything.

• High levels of competition in the marketplace

• Relatively low cost of predictive analytics software.

• Necessity of  providing good decision support tools to as many as possible across the enterprise

As far as the IT architecture is concerned, predictive analytics is considered to be a sub-operation within Business Intelligence framework. The following diagram illustrates how different parts of BI sub systems relate:

Query/Reporting and OLAP (On Line Analytical Processing) operations are directed towards the past whereas predictive analytics looks at the future.

Benefits of predictive analytics can be huge.  Certain things must be in good shape before one can realize the benefits.

1. Data must be of good quality. Garbage-in Garbage-out is more prominent in predictive analytics than many other operations.

2. The actual predictive analytics tool(s) used.

When the right things are in place (isn’t it with everything?) ROI of predictive analytics is always positive and large. Numbers like 145% ROI (IDC Report) in the first year are common place.

The reason for strong ROI is the fact that predictive analytics can and will discover hidden information that is already in your data and having that additional information can improve revenues or reduce costs dramatically. Today, most common uses of predictive analytics is in banking, insurance and consumer product industries (CPG and Retail). Other sectors have taken note and they are coming on stream with speed.

A good example is healthcare. Being notoriously inefficient, healthcare has tremendous advantage for improvement through the utilization of predictive analytics. Some examples would be to forecast future healthcare costs accurately, scheduling staff and other resources as well as diagnosis of patients for relatively common diseases like diabetes where there is enough quality data. 

Clear trends driving the growing demand for predictive analytical tools and capabilities:

• Increased need for predictive analytics. OLAP-based business intelligence applications lack the sophistication required to deliver the predictive analytics companies need to be competitive.
• Ever-increasing data volumes. As the amount of information that companies collect grows, the expense of storing data increases, placing additional pressure on companies to ensure that they extract value from the information they amass.
• Proliferation of methodological research. The imperative to discern trends and opportunities rapidly is making predictive analytics a more mainstream component of enterprise business intelligence.
• The power of visual information. As data volumes and data complexity increase, companies are beginning to recognize the value and benefit of being able to communicate complicated numeric information using well-designed charts rather than simple tables.
• Regulatory requirements for statistical analysis. New regulatory mandates require companies to employ advanced predictive analytic methods to achieve compliance.