Navigating Predictive Analytics: Benefits and Challenges of Using Graph Theoretic Methods Leveraging Contextual Network Information for Better Accuracy

Nick Lim is the CEO and Founder of Sonamine, LLC. Over the past 15 years, Nick has worked in analytic environments with large amounts of data. Leveraging his training at Harvard, he has led product and strategy teams at MicroStrategy, Enpocket and Nokia to build systems that predicted ad clicks and consumer behavior using terabytes of information.

Predictive Analytics: Benefits and Challenges of Using Graph Theoretic Methods Leveraging Contextual Network Information for Better Accuracy

The practice of predictive analytics has come a long way since the advent of operations research. Current predictive models routinely use more than 100 variables, most of them are characteristics of the object being analyzed, including age and gender for customer analysis, or color and text for advertisements. In this article, we will explore how graph methods can be used in predictive analytics.

Graph methods refer to techniques that analyze a network of objects, rather than pie charts and trend lines. Networks are made up of nodes and connections among the nodes. Other synonyms include vertices, edges, and arcs. Graph methods analyze the link structure, rate of diffusion and the clustering of nodes in a network. Many concepts in the popular press such as Google’s page rank, social network analysis, influencer marketing and driving directions all utilize graph theoretic methods.
Real World Use Cases
Why are these graph methods useful for predictive analytics? One reason is that additional contextual predictors can be derived using graph methods. Let’s start with a well known example: Google uses page rank as one of many predictors for the relevance of a web page. The link structure in the world-wide-web network provides valuable contextual information about which pages are deemed most relevant by the web page creators—this contextual link structure is then used to predict relevance for a user’s query.

Another example: If you are trying to predict the revenue contribution of each employee, in addition to the educational level, job responsibility and experiences, you might consider how well the employee relates to the customers or other managers. The contextual information added here is not something inherent about the employee. IBM conducted such a study with 2600 of their consultants, and created a model with revenue as the outcome target variable and three sets of predictor variables.1 In addition to the standard employee and job-related characteristics, two sets of contextual variables were added. One was related to the structure of the web of relationships, and the other was related to how each employee was related to managers and customers. It turned out that graph-derived contextual predictors were highly correlated with revenue.
Enriching Context Where It Seemingly Did Not Exist
One common question raised is how much of the contextual information is already captured in the demographic data. After all, there is a common understanding that birds of the same feather flock together. To tackle this question, a group of researchers at MIT studied how the Yahoo instant messenger network affected the adoption of the Yahoo Go product.2 They discovered that only 50% of the contextual network effect could be explained by the common underlying demographic variables.

This finding has big implications for predictive analytics in data poor environments. A data poor environment would benefit greatly from using contextual predictors that you already have. For example, if you are trying to predict churn in a prepaid mobile customer base where there is little to no demographic information, using the social network contextual information can greatly improve the prediction accuracy and lower the false positive rate.

Another type of contextual network information that can be included into predictive analytics is the underlying form of the social community that one belongs to. In the study with Yahoo data above that was trying to predict product adoption, they found that the chances of you adopting a product rises if more people in your social community have already adopted the product.

Besides social contextual network data that can be added to predictive functions, you can also deduce contextual information. For example, in trying to predict fraud from a set of credit applications, we can construct a network of information attributes. Using the basic idea of that people with bad credit tended to relate to other similar people, some researchers generated the graph-based hub and authority variables from such a network and added them to a support vector machine prediction function.3 The graph-derived context variables improved the fraud prediction by approximately 20%.

Clearly, the benefits of adding graph-based predictors are better lifts and lower rates of false positives. The improvements range from 20% to 100% depending on the richness of data available today. These benefits translate directly into bottom line results to the extent that your company is already putting predictive analytics into daily operations. So what are the practical aspects of putting this into operation?
How Do You Get Started?
Let’s start with the data: Where do you obtain the data to construct such contextual networks? The IBM study looked at the anonymized email traffic, address books and buddy lists. If you were a retailer like Amazon, you could use wish lists, product gift transactions and share-the-love data to construct the social context for predictions. Financial services companies can use the trading data to construct the contextual network among the traders or credit application data to construct an information network. Telecommunications companies can leverage the call detail records. In fact, the explosion of data collection means that most organizations already have this contextual network information just waiting to be accessed and utilized.

One unique aspect of contextual network predictors is that they are usually specific to the target outcome you want to predict. This is the definition of contextual after all. So if you want to predict churn, then you need to understand how the contextual network predisposes a subscriber to churn. Such contextual churn insight will certainly be different if you were trying to predict employee revenue or product adoption.

Since the contextual environment is changing, the window that an enterprise has to leverage the additional predictive power of contextual networks is also dynamic. Most customers will switch vendors very soon after they learn about a bad experience from a friend, so vendors must respond in a timely manner. Practically, this means that the calculation of graph-based predictors must occur more frequently.

Some contextual network information is relatively stable over time. For example, if someone has a preferred takeout restaurant, she will likely continue to call that restaurant even after she gets a new phone number. These “creatures of habits” can be more easily identified, say in the case of fraud or terrorism prevention.
Challenges
One potential impediment to leveraging contextual network information involves the privacy and data security issues. Since contextual networks tend to involve personal transactions, companies would like to ensure the highest security to prevent breaches. Various regulatory bodies at different state, country or global levels might have conflicting guidelines on what information is considered private and cannot be used for data analytics.

Another obstacle today is the scalability challenge. Often these contextual networks are many times the size of the current data used for analysis. Contextual networks rely on relationship information, so sampling runs the risk of losing valuable information. James Kobelius of Forrester postulates that the advent of social network analysis will catalyze the need for petabyte-sized data warehouses.4

Another obstacle is the lack of software tools that will generate these graph-based predictors. Types of specialized graph based calculations include community detection, diffusion simulation, vertex similarity, topology and centrality analysis. Open source tools such as Jung and Pajek work well for R&D, while proven commercial tools such as Sonamine are geared toward large scale and production graph mining.

A last obstacle is education. Contextual network predictors are unlike other types of data being captured and analyzed. Enterprises must first capture this new data, and that will require some mind-set shift. Setting up an infrastructure to model the contextual network will require new tools and approaches. Finally, using these contextual networks in predictive analytics that improve the bottom line will require the cooperation of multiple disciplines—analytics, marketing and operations.

Despite obstacles, the time to investigate and leverage these graph-based methods is now. IDAnalytics, a company that provides identity theft protection, has been issued a patent for a system and method for identity-based fraud detection through graph anomaly detection. How will you start adding graph-derived predictors into your predictive analytics?

David Katz has been in the forefront of applying statistical models and database technology to marketing problems since 1980. He holds a Master’s Degree in Mathematics from the University of California, Berkeley. He is one of the founders of Abacus Direct Marketing and was previously the Director of Database Development for Williams-Sonoma.

Check out the three minute interview on the Predictive Analyticswebsite.

Transcript:

Bill Cullifer: David Katz, Founder and President David Katz Consulting in Oregon. Good morning Mr. Katz and thanks for agreeing to the call.

David Katz: Well, thanks for having me.

Bill Cullifer: Mr. Katz, you are a veteran of the statistical data model and database technology market and I am curious if you could share with the subscribers of this podcast an historical perspective of predictive analytics?

David Katz: Well, sure. You know looking back early days of you know computers in business was all about cost control or cost reduction. You know, people would do applications primarily based upon the idea that they are going to get more work done and use fewer resources too. Over time, with the growth of processing power and storing technology, this… the perspective gradually shifted. You know, the first step was more of a reporting, business reporting; you know getting more information about what is happening in the business and then a further evolution comes in to the idea of predictive analytics or any kind of analytical approach through the database that is developed. So, now the focus is much more on serving the customer better, getting strategic insight, and being able to be agile in the market.

Bill Cullifer: Yeah, great summary and historical perspective. Where are we today in terms of the opportunity in the market? Is this something that enterprise understands completely and very well and are taking advantage of it, is it on the other hand going mainstream?

David Katz: Well, I think people have a lot more awareness than ever before that there is opportunities there. I mean people understand that there are tools and what I see is really a wide disparity from one company to another and from industry to another. Obviously, the people who are involved in Internet technology or engineering, they tend to be much more savvy about the possibilities of predictive analytics. Direct marketers have been really in the forefront of it, but you know there is a lot of areas where they are just waking up to this healthcare and education. You know there is a lot of these industries are just at the point where they need to standardize the information, to make it more available for analysis. So, there is a wide variety of places in different industries and different companies they are at[.

Bill Cullifer: Thanks. I appreciate that perspective. If I were a CEO of a mid sized company today, what would you like me to know about predictive analytics?

David Katz: Well, really you know the heart of it is, you know, most reporting for businesses are for finding purposes. You can only look at a few variables at a time. Predictive analytics really comes into its own when you have you know a volume of data with many interacting factors that are kind of hard to tease out without the right tool. So, you know, you find that a lot of times you have many, many factors that might be associated with the outcomes that you want, but they are all correlated with each other and so you need some expertise and some tools to help you make the most of that data and that’s just kind of a rough introduction, but I think that you know the key thing is when there is more data that you can take in, in some simple report, that’s when you want to start thinking about more sophisticated tools.

Bill Cullifer: Excellent. Well, I certainly appreciate the information and your time today Mr. Katz.

David Katz: Well, it’s been a pleasure. Thank you very much.

Mr. Katz is a visionary and thought leader in the emerging Marketing Operations field. He is a veteran with more than twenty years of marketing and change management experience in the technology industry in corporate, agency and entrepreneurial positions, where he directed corporate marketing, communications, public relations, lead generation and qualification, investor relations, and employee communications programs.

What you will learn:

* Marketing Operations::The Engine Behind Predictive Analytics

The holistic application of predictive analytics to create a smarter, more agile enterprise is unquestionably attractive, even sexy. But predictive analytics cannot fulfill its considerable promise without a strong Marketing Operations (MO) organization behind it to give it direction, meaning and opportunities for practical application.

What is Marketing Operations?

A Definition:

Marketing Operations is a thorough, end-to-end operational discipline that leverages processes, technology, guidance and metrics to run the Marketing function as a profit center and fully-accountable business. It reinforces Marketing strategy and tactics with a scalable and sustainable enabling infrastructure, as well as nurturing a healthy, collaborative ecosystem, both within and outside the Marketing department, to drive achievement of enterprise objectives.

Transcript:

Bill Cullifer: The mobile phone Gary Katz founder and CEO of Marketing Operation Partners good afternoon Gary and thanks for being in the call.

Interviewee: Welcome.

Gary Katz: Gary you have given a presentation [Inaudible] conference, the title was the engine behind predictive analytics can you tell us a little bit about that?

Interviewee: Yeah, basically my focus is on emerging the older discipline of marketing operation and basically how bad it is now and the effectiveness of marketing operations within the enterprise is to see how effective predictive analytics will apply.

Bill Cullifer: Fair enough, could you expand on that a little bit.

Gary Katz: Sure, well marketing operation is increasingly becoming a function that organizations with the complex marketing requirements and that are spending a fair amount of money in marketing in a lot of marketing [Inaudible] in order to run the marketing function more like a fully accountable business profit on value [Inaudible] but basically what marketing operation does is leverage profit from the technology [Inaudible] giving the chief marketing officer effectively, a chief of staffing make sure that the operational issues [Inaudible] discipline and also that marketing plays a role in [Inaudible] and it change have [Inaudible].

Bill Cullifer: What would be a couple of the higher level walk away?

Gary Katz: Absolutely that is a good question. Well, it depends on the you know what the [Inaudible] role is obviously, but you know at the C level it in not only embrace marketing operations and they have it all ready, but if they are already doing marketing operations then do it more broadly and more of a holistic strategic and operational approach, but make sure that the execution of the marketing is up as effective as the strategy and the vision that they come up with firstly if there is good alignment between the two and if there is good alignment in the marketing function and other [Inaudible] functional organization that [Inaudible] play a key role whether it is marketing efforts.

Bill Cullifer: Is there a technical component to this or is it all strategy?

Gary Katz: Well, there is always a technical component, but I am not diving into the nut bolts and you know I don’t think I [Inaudible] against that [Inaudible] and so…but marketing operation is the way to lay the ground work, so that the infrastructure [Inaudible] and the different approaches work well, so whether that infrastructure happens to be in a way a predictive analytics approach that includes both subject matter expertise and some kind of technology [Inaudible] but also enterprise marketing management type of technology [Inaudible] technology digital asset management technology and management technology [Inaudible] so marketing operations basically is that it [Inaudible] both technology initiative to make sure that they are that technology adopted are within the organization understood utilized [Inaudible].

Bill Cullifer: Fair enough, well thank you for your time today Gary excellent summary.

Gary Katz: My pleasure Bill. Analytic is a very [Inaudible] area and I think it you will have a lot to do with the [Inaudible] marketing to be more scientific more predictable, more consistent which is the kind of thing we are looking for in order to prove the operational performance of marketing and thus have a bigger impact on the [Inaudible] of an organization to accomplish this.

Bill Cullifer: Very well said.

Today’s topic is Predictive Analytics and this is the first in a series of audio podcast interviews on this topic. Dr. Eric Siegel is a former computer science professor at Columbia University and the President of Prediction Impact.

Here are questions that I asked:

Q: What is Predictive Analytics and does it accomplish?

A: Predictive Analytics is business intelligence technology that produces predictive scores for each individual customer or prospect. What you need to do before employing predictive analytics, is first decide which customer behavior will be most valuable to predict, such as predicting which customer is most likely to respond to an offer or which customer is most likely to cancel their subscription. And the next thing you need to do is prepare the data. Your data, which is, essentially, your organization’s collective experience, is leveraged by predictive analytics to produce predictive models and in so doing you’re actually learning from experience.

Q:  What kind of investment in infrastructure required?

A: Well you can start with a pilot initiative for very little, in the way of hardware and software requirements.  And this is also, you know, a place to start, to achieve a proof-of-principle, demonstrating what kind of return-on-investment can be achieved, such as improved customer retention, or increased profitability of a campaign.  In this case, the core predictive modeling can usually be done with free evaluation software licenses or with a free open-source tool.

Having said that, however, it’s important to note that what you do need is expertise in predictive analytics, either by way of internal resources, and/or employing professional services.  This expertise is needed to optimally position this technology, in order to determine the kind of behavior that’s going to be most valuable to predict on the business side, and then, more technically, what data’s require to achieve that prediction goal, and how you really need to prepare that existing data you have now in the right form, so that the resulting predictions you end up getting will be valuable in that they’re accurate and business-actionable.  And then you apply what’s learned by way of directing, let’s say, a retention campaign, or selecting targeted content on a per-customer basis as far as what’s the product or message each customer is most likely to respond to.

Please follow the link to http://predictiveanalytics.org/?p=6  for the podcast audio interview.