Predictive Analytics in the Real World: What Does It Look Like?
The Global Analytics Product Marketing Manager for Dell Statistica explains predictive analytics' biggest strengths, and how companies can leverage them.
Predictive analytics isn’t a brand-new technology, but it is one that has just started to come into its own in recent years. Thanks to advances in how big data can now be collected and handled, exploring that data and using predictive analytics is moving within reach of more organizations than ever before. But what does that really mean for the organizations who are using it? We talked to David Sweenor, the Global Analytics Product Manager for Dell Statistica, a predictive analytics software designed to make data analytics faster, more accessible and more usable for business.
Techopedia: Can you explain a bit about what predictive analytics can do for a business, and what it takes to move beyond analyzing data and predicting future results to actually translating that information into action?
David Sweenor: Statistica has been around for over 30 years as a predictive analytics platform that's deployed in all industries. I'll give you a few examples of what it can do for a business. One of our customers in Mexico provides micro-loans. If a user wants to apply for credit, they go to a website, enter their information, and a predictive model delivers a real-time score that determines whether they should be given a loan. This is important because in many parts of the world, traditional credit bureaus like FICO, Experian and Equifax are either non-existent or unreliable. Additionally, the banking laws also differ so the company can supplement some of their more traditional data with social media data for example, and can create a predictive model that provides a better risk profile of the applicant. In doing so, the company was able to reduce their default rates by over 80 percent. That’s game-changing for a lender and it’s something that’s not possible if you’re not connecting to and analyzing all the data that’s available. That's just one of the many examples we have in the banking world.
Embed Analytics Everywhere: Enabling the Citizen Data Scientist
Let’s look at healthcare for another example of a company using predictive analytics to fundamentally alter the way they do business and interact with clients – in this case, patients. Within the healthcare field, sepsis and surgical site infections are really a big problem. It's the number one preventable problem in hospitals today. At the University of Iowa Hospitals and Clinics, they’re using predictive analytics with Statistica to address the problem head on. Statistica combines historical patient data with real-time data obtained during surgery itself, and right there, in real time as the surgery is ongoing, gives the surgeon a predictive score to determine whether that patient is likely to develop a surgical site infection.
If that score is above a certain threshold, the surgeon can take preventative action right there, within the operating theater before the surgery has concluded. The results have been nothing short of staggering – a 74-percent reduction in the occurrence of SSIs for patients of colon surgery during the three-year period ending December 2015.
Those are just a couple of examples of how predictive analytics can be used, but it’s something companies in all industries can leverage. We have a large presence in manufacturing, for example. Regardless of what industry you’re in, predictive analytics allows you to make fact-based decisions to improve operations, improve safety, improve health and reduce risks.
Techopedia: What are some of the key big data problems predictive analytics software is designed to solve?
David Sweenor: I think one of the key problems is that our ability to collect data will always outstrip our ability to analyze the data. So, there are a couple of things that are happening within the industry because of that. The historical way of analyzing data would be to collect data from different systems and bring this massive volume back to a centralized server or a data warehouse of some sort. Then, you would have some specialists – data scientists, statisticians, mathematicians – who would analyze this data in a basement. They would eventually emerge with some answer to help the business move forward.
That's problematic in that we no longer have nine months to wait for data to be analyzed. We need insight and answers in real time – at the speed of business.
What we need to do then is give people within an organization some sort of a self-service capability. And that’s exactly what we’re focused on with Statsitica. We have technologies within our platform that allow data scientists within the organization to create an analytic workflow. They can store that workflow in a central repository, and then some other person who is not as skilled can take that workflow and reuse it. It's a node, it's an icon – they drag it to their canvas and use it in their own workflow as if they’ve built it themselves. Now, maybe that reusable workflow template is very sophisticated and does some advanced modeling that they don’t fully understand, but now that someone has built an advanced analytic workflow that can be put to work on what they do understand – their data and their business.
So, really what we're trying to do is help people stop reinventing the wheel and doing things over and over. That’s people side. On the technology side, what we’re focused on is making it so that rather than taking all this data and shoving it back to a central server for analysis on the data side, we can bring the analytics to the data.
In the end, if the analytics can live wherever data is – it could be in a database, it could be in Hadoop, it could be on a device, or a sensor within a refrigerator, or a light bulb, or wherever you can collect data – then we can save the time, money and energy we’re spending today moving data to and fro, and in the process make our analytics better, faster, more secure and more reliable.
Techopedia: Part of the role of predictive analytic software is to help empower what we call the Citizen Data Scientist. Why is this so important in the industry today?
David Sweenor: I think it comes down to a skill shortage. And when it comes to citizen data scientists, many of them are using Excel to manipulate data. That's great, but you can only go so far with Excel – they need something else. They need the ability to take data from disparate and multiple systems, internal and external to their organization, and combine it all together. They need automated ways to cleanse and prepare that data. They might know what kind of analysis they need to do, or what problem they're trying to solve, but they may not be familiar with the intricacies of how a neural network actually works, so that's where the data scientists would create a reusable workflow template that uses a neural network, or decision tree, or some other algorithm.
Me, as a citizen data scientist, I don't have to know the details of how the algorithm works, but I know what problem I'm trying to solve, and I have the ability to access to data within my sphere of influence. Statistica helps me do that by simplifying the process and abstracting the complexity so I can focus on solving the business challenge.
Techopedia: What are the most common business areas that predictive analytics is currently being used, and what areas is it expanding into?
David Sweenor: I would say analytics pervades every aspect of our lives, across all industries. We already talked about major advancement in the way healthcare and finance companies service customers thanks to predictive insights. We’re also now seeing customers use analytics to drive product innovation and to drive greater efficiency in the way they mold offerings to their clients’ likings.
For example, we work with a large organization that’s trying to come up with new offers and entirely new ways of doing business for different products the company offers.
Prior to using our technology, it would take them about nine months to do a proof of value, or a proof of concept in terms of a new way of doing business. After they implemented Statistica, that proof of value went from nine months to one day. That was an amazing result for them.
Techopedia: How should companies define objectives for business analytics model?
David Sweenor: I would start with clearly defining the business question you're trying to answer. You need to understand what levers to pull. The technology, I think, will always be there. It's best to start with a question.
So, start with that question or set of questions. What can you influence within your business? And then work backwards and model that decision. We don't want to just do predictive analytics for the sake of it, we want to be able to change a behavior.
Techopedia: Dell's Statistica has been praised as one of the leaders in the advanced analytic platform space. How is it separating itself from the pack, and solving the problem of creating actionable insights out of data?
David Sweenor: I think it's really the innovation that we’re constantly driving. We always listen to our customers when we develop the software. Companies realize that citizen data scientists and line-of-business users are the future of analytics. They want to use all those people and all the skills within their organization. We're making it easier for those people that are not experts in mathematics to be able to use math to arrive at an answer. Our environment is open. What that means to them is that their data scientists can code in a language-appropriate to whatever they want to achieve. It could be Python, it could be R, it could be anything, really. But it's easy to deploy, and for those who don't know how to code, like me, they can drag and drop icons to build a workflow and make a decision.
We're also enabling organizations to re-use the technology investments they have. So maybe they've invested in databases or Hadoop, or other things like that; we use those systems. We can do analytics directly within those source systems.
It's very innovative to be able to take an analytic workflow and apply it within different target environments. No longer am I bound to the four walls within my factory, or my hospital, or by whatever border have traditionally surrounded my business. We have technology that allows us to do this across sites and across geographies. It's the innovative component that is allowing us to step up and lead the market in this area.