Evaluate Weigh the pros and cons of technologies, products and projects you are considering.

Predictive analytics tools: How APIs, AI make data accessible

Apigee's Waqar Hasan discusses must-have features for predictive analytics tools and how to leverage the technology.

New predictive analytics (PA) tools, like Apigee Insights, are adding application program interface (API) and artificial intelligence(AI) to business intelligence (BI) technologies to enable proactive customer support and in-house development of analytics apps.

In old-school boardroom meetings, executives brainstormed about the next new thing, based only on sales charts and intuition. Today, predictive analytics tools can take the guesswork out of predicting what customers will need and want in the future. Until recently, only major companies like Google or Amazon could afford predictive analytics on "big data." New tools, like recently released Apigee Insights, will drive democratization of PA by enabling enterprise developers to build their own analytics applications, according to Waqar Hasan, CEO of InsightsOne, now part of Apigee.

Hasan recently worked with Independence Blue Cross (IBC) on a project to deliver proactive customer service. He describes the IBC project, the power of predictive analytics, must-have features for PA tools and uses for Apigee Insights in this interview.

Waqar HasanWaqar Hasan

What is driving the need for predictive analytics now?

Waqar Hasan: Predictive analytics as a concept has been around for a while but with limited adoption. Here's what changed in the past 10 years: Internet companies like Google, Yahoo and Amazon created the technologies to cheaply use big data to serve consumers better. They beat others to the ability to collect data, put it together -- mostly by using Hadoop -- and gained significant competitive advantage. The next step is for all enterprises to put their data to use, to increase revenue with it.

Which technology advances have made predictive analytics more accessible?

Hasan:  The three advances that have made predictive analytics both more affordable and more precise are plummeting hardware costs over the last decade, distributed data processing and machine learning. This are opening up predictive analytics for broader application.

The most important advance is machine learning, an AI (artificial intelligence) technology that permits computers to adaptively learn from big data without requiring programming. For example, machine learning can help a Telecom company precisely predict which customers are likely to churn when their contract expires by analysis of their usage and billing patterns combined with analysis of sentiment in call notes taken by a customer service representative.

Machine learning at scale requires marrying it with distributed data processing technology. Hadoop is the modern distributed data processing technology than scales on commodity hardware. It allows companies to collect, store and process big data more cost effectively than feasible through older RDBMS technologies.

What are the must-have features in a predictive analytics application set?

Hasan: The most important capabilities for creating predictive models are the ability to handle a variety of complex un-structured real-time data at scale and machine learning algorithms that provide significant increase in precision. For consumption of predictive models, it is important to support creation of predictive apps by JavaScript programmers.

What is the role of APIs in enabling predictive analytics?

Hasan: Bringing together predictive analytics with APIs gives businesses the ability to produce predictive apps that know their customers, understand what the customer needs and can deliver to the customer what they want before they know they want it.

We are seeing an increasing percentage of real-time customer interactions flow through APIs. This makes APIs a great place both to collect data as well as expose real-time predictive analytics to App developers.

At the end of the day it is the usage of predictive analytics that drives business value and we think API’s are the right place to drive adoption of real-time predictive analytics.

Could you offer an example of how predictive analytics adds value to business processes?

Hasan: Predictive analytics changes customer service from reactive to proactive. Look at customer service through call centers as an example. We worked with Independence Blue Cross, a major insurance company. Typically, the customer center reps type out information from a conversation with a customer, and that data goes into call center logs. Also, insurance claims are generated when customers visit their doctor or pharmacy, including information such as denied claims, copays and delays between visits. Using all this data, our predictive analytics tools are able to predict which customers are likely to be unhappy before they actually call Independence Blue Cross.

The next step is for all enterprises to put their data to use, to increase revenue with it.

Waqar Hasan,
CEO, InsightsOne

Knowing who might be dissatisfied, customer service reps can proactively reach out every day to a very small number of customers and say, 'We understand something might not be just right for you, and let's make it right.'

This kind of use case can apply to many businesses' relationships with their customers. Predictive analytics brings some tangible business benefits from gathering and storing big data.

What technologies work together to get this outcome for IBC?

Hasan: Machine learning on big data comes into play. In the IBC project, Apigee Insights was used to combine unstructured data in the form of notes with complex structured data in the form of claims in real time. Further, as more data is available, the machine learning automatically adapts, thus making the outcomes much more precise.

Machine learning enables conclusion about what will happen to be drawn from these powerful collections. This is game-changing. It is now possible to create business value from it in areas that were not possible before. That is not possible with classic statistical modeling, which does not handle unstructured data, complex structured data or real time.

[Note: Apigee Insights comes in several offering tiers, and the enterprise version starts at $120,000.]

How does Apigee Insights differ from Hadoop distribution providers?

Hasan: Apigee Insights is complementary to Hadoop providers such as Cloudera and Hortonworks. We help customers unlock business value from their Hadoop investment. Suppose somebody has brought Cloudera to collect data into a data lake. Next, they want to do some predictive analytics to impact their business. Apigee Insights helps them create predictive models as well as expose the predictive intelligence so that app developers can deliver predictive apps.

Apigee Insights is complementary to Hadoop providers by helping customers to create value on top of those platforms. Also, we are complementary to business intelligence solutions. Apigee Insights provides predictive intelligence, combining APIs with predictive analytics to deliver the ability to create predictive apps.

What's your advice to enterprise architects in working with the business side on getting started with predictive analytics?

Hasan: Start with a problem which, when solved, has business value to the company and for which data is available. Make sure there is a clean business case.

Also, do not think along classic enterprise architecture lines. In classic enterprise approaches, you first identify the business use cases and only then start constructing the data warehouse for those use cases. The new way with Hadoop is to get the data first and put it all together in a data lake. Once you have data in a data lake, it actually becomes much easier and faster for people to identify valuable use cases. This is how you unleash innovation.

Jan Stafford plans and oversees strategy and operations for TechTarget's Application Development Media Group. She has covered the computer industry for the last 20-plus years, writing about everything from personal computers to operating systems to server virtualization to application development.

Next Steps

How machine learning and predictive analytics help build towards AI

This was last published in May 2014

Dig Deeper on API management strategy

Join the conversation

1 comment

Send me notifications when other members comment.

Please create a username to comment.

Do you use predictive analytics tools?