Archive for July, 2016

July 29th 2016

Wise Practitioner – Predictive Analytics Interview Series: Gary Neights at Elemica

By: Eric Siegel, Founder, Predictive Analytics World

In anticipation of his upcoming conference presentation, Predicting Behavior in Chemical Industry Supply Chains, at Predictive Analytics World New York, October 23-27, 2016, we asked Gary Neights, Senior Director at Elemica, a few questions about his work in Gary Neights IMAGEpredictive analytics.

Q: What are the challenges in translating the lessons of predictive analytics from other verticals into manufacturing?

A: Predictive models for pharma and retail are often used to influence consumer behavior or give direction to research efforts. These models yield results that are reviewed by experts before being acted upon.   The predictive system for manufacturing supply chains that I will discuss drive real-time manufacturing execution decisions by front-line employees.  Commitment of resources such as labor, manufacturing capacity, raw materials, and logistics capacity can occur in near real-time.   Accurate, real-time data flows from customers, distributors, suppliers, and carriers are required to develop a full picture of the situation and help provide the best level of decision support.

Q: In your work with predictive analytics, what behavior do your models predict?

A: One example is under supply or over supply conditions.  Over supplying finished goods may lead to price discounting while under supplying material to a downstream manufacturing process may shut down operations.  Another example is predicting which perishable materials in a complex supply chain network are nearing expiration so they can be expedited to an appropriate manufacturing facility.

Q: How does predictive analytics deliver value at your organization? What is one specific way in which it actively drives decisions?

A: Supply chain decisions have financial impact and need to be taken in near real-time.   Product-by-product and plant-by-plant predictions can lead to information overload and indecision.   For example, if rail cars to a manufacturing site are predicted to be late do I dispatch trucks as a rush shipments… or dip into safety stock?   If trucks, how many?  Over the long-term data may be analyzed systematically and accounted for during periodic planning cycles or contract renegotiations.

Q: Can you describe a successful result, such as the predictive lift (or accuracy) of your model or the ROI of an analytics initiative?

A: A predictive Railcar KanBan system drove a working capital savings of greater than $400K / year for one product and inventory replenishment accuracy was increased from less than 55% accuracy to greater than 80%. This allowed a 20% reduction in safety stock levels and 40% reduction of leased railcars. 

Q: What surprising discovery have you unearthed in your data?

A: In one case a graphical review of time-series data indicated that a manual supply chain management process was systematically driving costly inventory swings.  The planner was not correctly accounting for transit times between locations, nor the operating hours for shipping and receiving operations. This was corrected by a correctly tuned predictive system.

Q: Sneak preview: Please tell us a take-away that you will provide during your talk at Predictive Analytics World.

A: A common theme I hear is that the farther you are from the consumer the harder it is to get accurate demand data.  We will share one approach that supports manufacturers systematically aggregating demand to improve predictive accuracy.

———————

Don't miss Gary’s conference presentation, Predicting Behavior in Chemical Industry Supply Chains, and workshops at Predictive Analytics World New York on Tuesday, October 25, 2016 from 3:05 to 3:25 pm.  Click here to register for attendance.

By: Eric Siegel, Founder, Predictive Analytics World

No Comments yet »

July 27th 2016

Wise Practitioner – Predictive Analytics Interview Series: Dr. Sarmila Basu at Microsoft Corporation

In anticipation of her upcoming conference presentation, Predictive & Prescriptive Analytics Helps Keep Kids in School at Predictive Analytics World London, October 12-13, 2016, we asked Dr. Sarmila Basu, Senior Director, Data & Decision Sciences Sarmila Basu IMAGEGroup at Microsoft Corporation, a few questions about her work in predictive analytics.

Q: In your work with predictive analytics, what behavior or outcome do your models predict?

A: We do predictive analytics for business groups within the company as well as for customers outside. For internal groups our predictive analytics involves things like:-

  1. Predictive models for customer attrition for a specific subscription based product.
  2. Models for antipiracy efforts to identify compromised license keys
  3. For some of our external customers we have built models for predictive maintenance to plan for potential failure of their machineries.

 

Q: How does predictive analytics deliver value at your organization – what is one specific way in which it actively drives decisions or operations?

A: Predictive analytics has delivered impact through revenue enablement and cost savings. When we are able to identify pirated/compromised license keys, that helps with revenue recovery. In cases of customer attrition modeling, we were able to start proactive retention efforts.  

Q: Can you describe a quantitative result, such as the predictive lift of your model or the ROI of an analytics initiative?

A: Our antipiracy analytic effort has led to revenue recovery worth $130million last year.

Q: What surprising discovery or insight have you unearthed in your data?

A: Surprising discovery varies from project to project. We sometimes find unexpected information or insight by combining multiple data sources. People often complain about their data being inadequate for any meaningful conclusion, but it is more an issue of broken data than bad data. Data Scientists have to be able to stich data from multiple sources together and build a cohesive story.

Q: Sneak preview: Please tell us a take-away that you will provide during your talk at Predictive Analytics World.

A: I will be talking about some of the use case scenario that has led not only to revenue impact for us but more importantly in terms of social good, it has delivered great impact.

———————

Don't miss Dr. Basu’s conference presentation, Predictive & Prescriptive Analytics Helps Keep Kids in School on Thursday, October 13, 2016 at 11:45 am at Predictive Analytics World London. Click here to register to attend.

No Comments yet »

July 22nd 2016

Wise Practitioner – Predictive Analytics Interview Series: Dae Park and Vijay D’Souza at Government Accountability Office (GAO)

By: Sean Robinson, Program Chair, Predictive Analytics World for Government

In anticipation of their upcoming conference co-presentation, Characteristics for Those Dae Park IMAGEClaiming Social Security Benefits Early, at Predictive Analytics World for Government, October 17-20, 2016, we asked Dae Park, Assistant Director at Government Accountability Office (GAO), and Vijay D’Souza, Director at Government Accountability Office (GAO), a few questions about their work in predictive analytics.

Q:  How would you characterize your agency's current and/or planned use of predictive analytics?  What is one specific way in which predictive Vijay D'Souza IMAGEanalytics actively drives decisions in your agency?

A: A core function of the Government Accountability Office (GAO) is to use data to objectively evaluate government programs and spending. We both develop our own models and evaluate agencies’ own models to evaluate the effectiveness of government programs. For example, we have completed numerous evaluations of Medicare payment policies to evaluate possible effects on quality and access to care.

Q:  Can you describe the challenges you face or have already overcome in establishing a data-driven environment in your agency?

A: The biggest problem we face is acquiring and validating data from the agencies we audit. Data are provided in multiple formats and often have quality issues. These include duplicate, incomplete, or inaccurate entries, and inconsistencies among records. This challenge is compounded when we merge data from multiple agencies or multiple sources to conduct more sophisticated analyses.

Q:  Can you discuss any near term goals you have for improving your agency's use of predictive analytics?

A: There is an increasing awareness across our analyst community of the value of predictive analytics to measure the effectiveness of federal programs. This includes training and outreach efforts from our analytics experts to other agency staff. We are also taking greater advantage of open source tools for analytics and developing strategies to increase the amount of staff that can help with the initial stages of data acquisition, evaluation, and preparation.

Q:  Can you describe a successful result from the employment of predictive analytics in your agency, i.e., cost avoidance, funds recovered, improved efficiency, etc.

A: We analyzed TSA’s program to detect more risky air travel passengers and determined it was not based on sound evidence. In another case, we evaluated a Department of Transportation program that used a large amount of data to calculate safety scores for commercial trucking carriers and to ostensibly determine the likelihood of crashes. We found that violations used in the agency model either occurred too infrequently to be useful, or were not otherwise predictive of crashes.

Q:  Sneak preview: Please tell us a take-away that you will provide during your talk at Predictive Analytics World for Government.

A: To better understand the circumstances faced by those who claim early Social Security benefits, GAO examined demographic and occupational characteristics associated with early claiming. More specifically, we examined the characteristics and income of early claimers using data from the Health and Retirement Study. I look forward to discussing our findings in detail at the session.

———————

Don't miss Dae and Vijay’s conference co-presentation, Characteristics for Those  Claiming Social Security Benefits Early, on Monday, October 17, 2016 from 11:25 am to 12:10 pm at Predictive Analytics World for Government. Click here to register to attend.

By: Sean Robinson, Program Chair, Predictive Analytics World for Government

No Comments yet »

July 19th 2016

Wise Practitioner – Predictive Analytics Interview Series: Dean Abbott of SmarterHQ

By: Eric Siegel, Founder, Predictive Analytics World

In anticipation of his upcoming conference presentation, The Revolution in Retail Customer Intelligence, at Predictive Analytics World New York, October 23-27, 2016, we asked Dean Abbott, Co-Founder and Chief Data Scientist of SmarterHQ, a few questions Dean Abbott imageabout his work in predictive analytics.

Q: In your work with predictive analytics, what behavior do your models predict?

A: I’ve built models that predict a wide variety of behaviors and patterns. A short list is provided here:

  • Customer behavior: response, churn, product up-sell and cross-sell, best marketing creative, days to next purchase, days to next visit;
  • Signals (radar): tank, truck column of tanks; (sonar): man-made vs. biologic;
  • Financial: fraud or suspicion of fraud, debt repayment period, debt repayment amount, insurance claim repayment likelihood, claim amount of repayment.

Q: How does predictive analytics deliver value at your organization? What is one specific way in which it actively drives decisions?

A:  I’ll speak to SmarterHQ, the company I’m Co-Founder of and Chief Data Scientist. There is no one specific way our predictive models drive decisions, but they are involved in the decision-making process in several ways, all related to selecting customers to promote to, whether that be selecting customers to send an email to, show a display ad, or present content on a page that is of greater interest to the customer.

Consider our models that predict the likelihood that someone will purchase a product during a visit to the company’s web site within 3 days. Each visitor is scored while they browse on the website and at the end of their session. The company now wants to create a new campaign to increase sales of a particular product by emailing them a promotion code with a 20% discount. If the customer is likely to purchase a product on the web site within 3 days, the models will exclude these customers from the email list; why take away margin from sales that are likely to occur anyway. Or what if a customer was very likely to purchase within 7 days last week but is no longer likely this week? This is a form of churn (but based on expected behavior, not actual behavior), and these customers could be given incentives to visit again.

Q: Can you describe a successful result, such as the predictive lift of your model or the ROI of an analytics initiative?

A: It is difficult to describe most of the results my models have generated because they are considered sensitive information for the company or government agency. I’ve had models in use by organizations for 10 years before they were refreshed. I’ve had another model so successful that it was put on the “do not tell” list by the organization because it became a strategic initiative for the organization. I’ve had fraud models identify multi-million dollar cases to investigate that were clearly fraud but had previously eluded detection.

Q: What surprising discovery have you unearthed in your data?

A: Most surprising? There have been many surprises over the years, usually related to the data itself and patterns of behavior that we may overlook, but are important nevertheless. For example, with the days to next purchase models, one expects that visitors on a website who look at lots of hot products are more likely to purchase soon; these are engaged visitors. However, it turns out that some of the most likely purchasers are those who visit just one item. The vast majority of the time, one-item visitors are not engaged and therefore are unlikely to purchase. But, if these one-item visitors were previously highly engaged, it’s a different story; they are focused like a laser beam on one product only. So the surprise was that there is this subset of visitors who look awful but are actually fantastic!

Q: Sneak preview: Please tell us a take-away that you will provide during your talk at Predictive Analytics World.

A: The most important take-away in my talk is this: When you prepare data for modeling, think about how the algorithms interpret the data. Each algorithm has weaknesses that can result in strange or misleading behavior. It’s our job as predictive modelers to help the algorithms do the best job they can.

Q: In addition to keynoting, you will be teaching two one-day workshops at PAW New York, Supercharging Prediction with Ensemble Models and Advanced Methods Hands-on: Predictive Modeling Techniques. How would you advise attendees to choose between these workshops and would it even make sense to attend both?

A: There are many workshop options, and all of them are worthy of attending. I think of the Supercharging and Advanced Methods workshops as complementary to your Online Introduction to predictive and John Elder’s Modeling Methods, with the sequence being (1) Intro, (2) Modeling Methods, (3) Advanced Methods, and (4) Supercharging. The Modeling Methods can be taken the day before Advanced Methods in the same conference; Modeling Methods provides a framework for predictive modeling, and Advanced Methods lets you try it out on commercial software. Supercharging takes predictive modeling to the next level, introducing the methods that win modeling competitions and have provided me with extra accuracy has made the difference between successful models and very successful models in my consulting practice.

———————

Don't miss Dean Abbott’s conference presentation, The Revolution in Retail Customer Intelligence, and workshops at Predictive Analytics World New York on Tuesday, October 25, 2016 from 10:30-11:15 amClick here to register for attendance.

By: Eric Siegel, Founder, Predictive Analytics World

No Comments yet »