May 17th 2016

Wise Practitioner – Text Analytics Interview Series: Emrah Budur at Garanti Technology

By: Steven Ramirez, Conference Co-Chair, Text Analytics World Chicago

In anticipation of his upcoming conference presentation, Tips and Tricks on Developing High-performance Fuzzy Name Search Engine to Prevent Terrorism Financing at Text Analytics World Chicago, June 21-22, 2016, we asked Emrah Budur, Senior Software Emrah Budur imageEngineer at Garanti Technology, a few questions about his work in text analytics.

Q: What is your topic mainly about?

A: Financial institutions and US businesses working with partners overseas must follow strict government regulations to prevent them from doing business with terrorist organizations or other sanctioned entities. In particular, banks are required to precisely detect inappropriate transactions with sanctioned entities out of large amounts of legitimate money transfers in real time. However, detecting names of sanctioned entities is challenging due to the high number of variations in terms of misspellings.

In this session, we will explore the tips and tricks of developing a high precision & recall fuzzy name search engine model that detects the names of sanctioned entities with greater accuracy and precision than search engine market leaders.

Q: In your application with text analytics, what behavior or outcome do your models predict?

A:  Our model detects the names of sanctioned entities out of free format input financial texts with a higher degree of accuracy than search engine market leaders. The final outcome is expected to be the names which exist fully or partially in the free format input text with exact or high similarity.

Q: Can you share a concrete example about the topic?

A:  Yes. When you search for “Barrak Obama” on Google (with the intentionally incorrect spelling) you will see a panel on the right hand side featuring some information related to Barrack Obama. This means Google applied fuzzy search on our query and figured out the exact identity we intended to seek. When you search instead for “Barrak Obama elections” on Google, the identity panel on the right hand side will disappear. Although the intended information that we seek is exactly the same, Google failed to show the identity panel on the second query. The banks are required to extract the intended identity included in both queries.

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

A: Text analytics is helpful in various decision making processes. For example, it can help you figure out the correlation between the number of false positive predictions and the level of similarity your model can tolerate, then make a data driven decision about how much similarity your model should tolerate to prevent excess number of false positives.

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

A: One of the most popular measures of success when evaluating search engines is the F1-Score (and also the F5-Score). F-Scores range from a scale of 0 to 1 (worst = 0, best = 1). We benchmarked our model with a hand-curated training dataset against the domain leaders. We achieved an F5-Score = 0.91 where the scores of domain leaders remained under .70.

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

A: We are surprised to find out some algorithms which are well known in some other domains are fairly applicable to search engine domain. For example, we were surprised to figure out that “the market basket analysis algorithm” can be applied to discover the frequent/infrequent co-occurrence of multiple stop words. For example, the terms “engineering”, “cooperation”, and “limited” can be considered to be stop words and may be filtered out from search queries because they return vast amounts of unnecessary information. And it may be possible that the statement “Cooperation Limited” is also frequent according to your dataset. However, the expression “Engineering Cooperation Limited” may refer to a unique organization. So, ignoring all of these stop words completely will lead you to a false negative match. On the other hand, identifying and unearthing the frequent co-occurrence of “Cooperation Limited” by means of market basket analysis will let you identify the importance of the term “Engineering” in this context hence come up with a true positive match instead.

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

A: Running name searches to prevent doing business with sanctioned entities in the global banking industry is a highly sophisticated and critical area in the field of text analytics. With the help of the tips and tricks that we will present in this session, a tailor-made solution can be implemented which provides more accurate and timely results than the solutions of leading search engine domains.

Q: Is it possible to share this case and comment or ask a question even before the session?

A: Yes! You are more than welcome to share your questions and comments on this Q&A platform and by #detectivenamesearch on Twitter.

———————

Don't miss Emrah’s conference presentation, Tips and Tricks on Developing High-performance Fuzzy Name Search Engine to Prevent Terrorism Financing on Wednesday, June 22, 2016 from 2:20 to 3:05 pm at Text Analytics World Chicago. Click here to register to attend

By: Steven Ramirez, CEO at Beyond the Arc, and Co-Chair of Text Analytics World

No Comments yet »

May 13th 2016

Wise Practitioner – Predictive Analytics Interview Series: Thomas Schleicher at National Consumer Panel

By: Eric Siegel, Founder, Predictive Analytics World

In anticipation of his upcoming conference presentation, Using Predictive Analytics to Optimize Organizational KPI's: A Panel Market Research Case Study at Predictive Analytics World Chicago, June 20-23, 2016, we asked Thomas Schleicher, Sr. Director, Measurement Science at National Consumer Panel, a few questions about his work in predictive analytics.

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

A: Our models predict a number of behaviors and outcomes, including attrition, compliance and the number of recruits needed to replenish households lost through churn. We also used predictive analytics in combination with test-control comparisons to further optimize business decisions.

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

A: At NCP, the ideal set of households participating on our market research panel would be fully representative of the nation's households while also fully participating in the panel by scanning and submitting all of their purchases. Using predictive analytics can help to ensure that the most optimal set of households is selected and maintained with respect to these and other metrics.

Q: Can you describe a quantitative result, such as the predictive lift of your model or the ROI of an analytics initiative? Do you see a role for integrating more “qualitative” impacts into a quantitative model?

A: We regularly evaluate ROI when we test new ways of engaging with our panelists in efforts to encourage them to remain as contributing panelists for years. However, we also take into consideration the value improved participation has to our clients. Although we are making progress on putting numbers on this sometimes qualitative impact, it is possible to undervalue an initiative by only referencing items that are easily quantifiable.

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

A: As we continue to develop our valuation of panelist compliance, we are learning additional ways to benchmark difficult to measure variables. More to come…

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

A: Although it can be challenging (or even impossible in some situations) to achieve “perfect” scores on your organization’s KPI’s, predictive analytics can leverage your data and business acumen to drive optimal business decisions.

———————

Don't miss Thomas’ conference presentation, Using Predictive Analytics to Optimize Organizational KPI's: A Panel Market Research Case Study on Tuesday, June 21, 2016 from 11:20 to 11:40 am at Predictive Analytics World Chicago. Click here to register to attend.

By: Eric Siegel, Founder, Predictive Analytics World

No Comments yet »

May 11th 2016

Wise Practitioner – Manufacturing Predictive Analytics Interview Series: Dr. Matteo Bellucci at General Electric

By: Bala Deshpande, Conference Co-Chair, Predictive Analytics World for Manufacturing 2016

In anticipation of his upcoming Predictive Analytics World for Manufacturing conference Matteo Bellucci imagekeynote presentation, Changing the Way we Make Things: The Brilliant Factory, we interviewed Dr. Matteo Bellucci, Manager, Process System Lab at General Electric. View the Q-and-A below for a glimpse of what’s in store at the PAW Manufacturing conference.

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

A: In a manufacturing environment such as the ones at GE, were we make low volume high value products, Return of Investment (ROI), R&D funding allocation and cultural resistance are probably the highest barrier for translating the lessons of predictive analytics into manufacturing. Given the “low volumes” it takes longer to see the benefit of an improvement on the factory floor. R&D is mainly dominated by Engineering functions that do not always fully understand the challenges of supply chain teams, thus allocating limited funding into new manufacturing technologies. Finally we still live in a world where ”trial and error” are dominating the mindset of supply chain teams, and the power of analytics is not fully engrained in the culture of those organizations.

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

A: The work we are doing at GE Global Research spans many different areas from virtually commissioning new factories before they are even built, to predictive maintenance algorithms or real time factory bottleneck detections.  

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

A: There are 3 main benefits of predictive analytics – 1) reduce cost of our supply chain operations, thus making GE more competitive; 2) increase the speed at which we can implement innovation in our products & supply chains; 3) drive real time decisions on the factory floor. For example if a CNC goes down unexpectedly, or if we see a surge in demand for a specific part, or if the production is “disrupted” by some R&D prototype that needs to be built, the system will inform us in real time of how to best utilize the assets that we have available within our internal or external supply chain, to still deliver at cost, and on time.

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: Over the last couple of years, we have proven many use cases in many of our factories. In one of them, a real time scheduling system is helping them gain significant productivity by providing suggestion on how best optimize assets on a day by day basis. Another use case was about predictive maintenance – we have proven to be able to detect an issue days before it would have actually occurred. We use analytics to improve parts yield and reduce cost of quality. We have seen many success stories where analytics has helped solve problems and deliver a better product, faster. Overall we believe the company is going to become stronger through the use of Brilliant Factory technologies.

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

A: There is an incredible amount of capacity at both machine level and plant level that can be unlocked by smart analytics that can help operators focusing on what really matters. Just by visually monitoring a machine you can get a 20% uptime increase. Most importantly the data are helping drive a behavioral change at the shop floor level and this is helping to accelerate some of the R&D and implementation efforts inside GE.

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

A: “If you went to bed last night as an industrial company, you’re going to wake up this morning as a software and analytics company.” – Jeff Immelt, GE Chairman & CEO. The presentation will be around Jeff’s quote with specific focus on how we are using analytics to optimize our own factories.

————————

Don't miss Dr. Bellucci’s keynote conference presentation, Changing the Way we Make Things: The Brilliant Factory, at PAW Manufacturing, on Wednesday, June 22, 2016 from 9:10 to 10:00 am. Click here to register for attendance. 

By: Bala Deshpande, Founder, Simafore and Conference Co-Chair of Predictive Analytics World for Manufacturing.

No Comments yet »

May 6th 2016

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

By: Bala Deshpande, Conference Co-Chair, Predictive Analytics World for Manufacturing 2016

In anticipation of his upcoming Predictive Analytics World for Manufacturing conference presentation, Analytics in Manufacturing Supply Chains – Predicting Behavior In Chemical Industry Supply Chains, we interviewed Gary Neights, Senior Director at Elemica. View the Gary Neigh imageQ-and-A below for a glimpse of what’s in store at the PAW Manufacturing conference.

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

A: Two things come to mind.  First is the number and variety of signals that need to be processed. Data from customers, distributors, suppliers, contract manufacturers, carriers, and third party warehouses may need to be analyzed to get the full picture.   Second is that predictive data may need to be acted upon quickly and decisions can commit resources such as manufacturing capacity, raw materials, or logistics capacity.  Decision support systems are critical.

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 shutdown 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: Product-by-product and plant-by-plant predictions can lead to information overload and indecision.   Supply chain decisions have financial impact and need to be taken in near real-time.  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: In one example the predictive system drove inventory replenishment accuracy from less than 55% accuracy to greater than 80%.    This allowed a 20% reduction in safety stock levels.  The number of leased railcars was reduced by 40%.  This drove a working capital savings of greater than $400K / year for one product.

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

A: The bullwhip effect can occur at a micro level.  In one case a supply chain planner was tightly controlling resupply – on the phone every day to make sure they never ran out of stock.  A change in today’s demand drove new truck shipment.  A graphical review made obviously that the manual planning process was systematically driving large inventory swings.  It did not correctly account for lead times as well as shipping, manufacturing, and receiving calendars.  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 for Manufacturing.

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, Analytics in Manufacturing Supply Chains – Predicting Behavior In Chemical Industry Supply Chains, at PAW Manufacturing, on Wednesday, June 22, 2016 from 2:15 to 3:00 pm. Click here to register for attendance. 

By: Bala Deshpande, Founder, Simafore and Conference Co-Chair of Predictive Analytics World for Manufacturing.

No Comments yet »

May 2nd 2016

Wise Practitioner – Manufacturing Predictive Analytics Interview Series: Jeffrey Banks at The Applied Research Laboratory at The Pennsylvania State University

By: Bala Deshpande, Conference Co-Chair, Predictive Analytics World for Manufacturing 2016

In anticipation of his upcoming Predictive Analytics World for Manufacturing conference presentation, Using Vehicle Digital Bus Data for Predicting Failure of Line Haul Trucks, we Jeffrey Banks imageinterviewed Jeffrey Banks, Department Head, Complex Systems Engineering & Monitoring at The Applied Research Laboratory at The Pennsylvania State University. View the Q-and-A below and glimpse what’s in store for the PAW Manufacturing conference.

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

A: The availability of high quality data is the key component of conducting useful predictive analytics.

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

A: Our work focuses on predicting electrical and mechanical component and system failures to reduce unscheduled maintenance and increase operational availability of critical assets.

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

A: We are helping the Department of Defense to implement predictive analytic solutions to enable them to transition to a data driven decision making organization.

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: We are still in the development phase of our work, so we do not have any impact results yet.

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

A: The big surprise is that though we are limited to using low sample rate time series data (~1 sample/second) we are able to demonstrate preliminary predictive results.

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

A: We will show preliminary predictive results for the work that we are doing for component and system failure for military ground vehicles.

————————

Don't miss Jeffrey’s conference presentation, Using Vehicle Digital Bus Data for Predicting Failure of Line Haul Trucks, at PAW Manufacturing, on Tuesday, June 21, 2016 from 11:20 am to 12:05 pm. Click here to register for attendance. 

By: Bala Deshpande, Founder, Simafore and Conference Co-Chair of Predictive Analytics World for Manufacturing.

No Comments yet »

April 28th 2016

Wise Practitioner – Text Analytics Interview Series: Frederick Guillot at Co-operators General Insurance Company

By: Steven Ramirez, Conference Co-Chair of Text Analytics World Chicago

In anticipation of his upcoming conference presentation, Leveraging Hands on Approaches to Identify Actionable Topics in Property Insurance at Text Analytics World Chicago, June 21-22, 2016, we asked Frédérick Guillot, Manager, Research and Innovation at Co-operators Frederick Guillot ImageGeneral Insurance Company, a few questions about his work in text analytics.

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

A: First of all, I’m working in the insurance world. As of now, our main focus had been to identify actionable topics in the free format text fields where our exclusive agents can write notes on everything they want about our clients. For instance, the most useful information we try to identify is the “life events” (newborn, wedding, new car, new home, new job, etc.) of our clients.

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

A: We are at a very early stage of using text analytics in the organization. For this moment, only research prototypes had been delivered. By the end of 2016, we will use text analytics to prioritize which clients to contact first, mostly for cross sell opportunities.

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

A: It is really difficult to articulate for this moment, but early tests in a recent marketing campaign showed an improvement by 15% on our cross sell acceptance ratio compared to a control group.

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

A: Hey, this was something we were really not looking for, but we discovered accidentally that a lot of our clients are annoyed by the fact that we have really strict underwriting rules for oil tanks older than 5 years. We are losing a lot of clients because of this rule, while 5 years of age is not necessarily too old for an oil tank.  

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

A: I will try to demonstrate that hands on efforts can have a great value while doing text analytics, especially in an industry heavily relaying in specialized terminology.  

———————

Don't miss Frédérick’s conference presentation, Leveraging Hands on Approaches to Identify Actionable Topics in Property Insurance on Wednesday June 22, 2016 at 9:20 to 9:40 am at Text Analytics World Chicago. Click here to register to attend.

By: Steven Ramirez, CEO at Beyond the Arc, and Co-Chair of Text Analytics World

 

No Comments yet »

April 25th 2016

Wise Practitioner – Predictive Analytics Interview Series: Alice Chung at Genentech

By: Eric Siegel, Founder, Predictive Analytics World

In anticipation of her upcoming conference co-presentation, Utilizing Advanced Analytics to Generate Insights at Predictive Analytics World Chicago, June 20-23, 2016, we asked Alice Alice Chung imageChung, Senior Manager at Genentech, 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 are trying to understand correlations between certain attributes that drive behavior changes of our study population. Understanding these relationships will help us address business questions and develop appropriate action plans to work effectively with our customers.

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 allows us to perform scenario planning to answer specific questions for our business. In both decision making and operational activities, we will be able to answer questions precisely and to anticipate outcomes based on market dynamics and business needs.

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

A: We are still working on collecting results and characterizing our predictive output. This is a work in progress for us at the moment.

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

A: It’s all about using and synthesizing data to answer specific questions for decision making. I have heard people hypothetically discuss various scenarios to utilize various predictive models yet there’s no data to use. I have also heard people spending time reviewing/analyzing data without business context or knowing what questions to address. Both approaches, in my mind, don’t yield meaningful insights.

For my team, we understand what and how our data can be used and the level of insights it can bring to support predictive models. The most critical component of our analysis and recommendations is to understand its limitations to answer business questions and identify methods to overcome them.

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

A: We will be discussing high level approach of our experiments and have dialogues with the audience to gather input and feedback based on their experiences!

———————

Don't miss Alice’s conference co-presentation, Utilizing Advanced Analytics to Generate Insights on Tuesday, June 21, 2016 from 11:45 am to 12:05 pm at Predictive Analytics World Chicago. Click here to register to attend.

By: Eric Siegel, Founder, Predictive Analytics World

1 Comment »

April 22nd 2016

Wise Practitioner – Manufacturing Predictive Analytics Interview Series: Carlos Cunha at Robert Bosch, LLC

By: Bala Deshpande, Conference Co-Chair, Predictive Analytics World for Manufacturing 2016

In anticipation of his upcoming Predictive Analytics World for Manufacturing conference presentation, Manufacturing Analytics at Scale:  Data Mining and Machine Learning inside Bosch, we interviewed Carlos Cunha, Senior Data Scientist at Robert Bosch, LLC. View the Carlos Cunha imageQ-and-A below to see how Carlos Cunha has incorporated predictive analytics into manufacturing at Robert Bosch, LLC. Also, glimpse what’s in store for the PAW Manufacturing conference.

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

A: One of the biggest challenges is in level of accuracy needed. In e-commerce and social networking applications, 80% accuracy might be very good and the consequences of errors are limited compared to that in manufacturing. Particularly, in manufacturing of safety critical applications, the accuracy requirements are stringent. Consequently, lot more complexity and sophistication goes into the provision of analytic solutions.

In root-cause analysis tasks, the challenge is to go from the correlations identified by the models to actual causation. The final proof can only be obtained by direct testing at the line the top potential factors.  But those tests can be time consuming and expensive for the plants, particularly if the plant has not yet fully embraced data mining methods.

Finally, verification and validation in manufacturing is an open challenge, under active research.

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

A: The target applications of our predictive models include the entire variety of business functions: manufacturing, supply chain and logistics, engineering, and Internet of Things and Services.    

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

A: Predictive analytics delivers value at all verticals of our organization; logistics, engineering, production, demand forecasting, etc. It guides our company, helping to decide what we build, how we build it, how to distribute our products and how and who to sell them to.

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: In some projects we have obtained up to a 65% reduction in scrap and up to 45% reduction in the time required for testing parts.

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

A: The best discoveries are the ones that are completely unexpected. It is very satisfying when our team discovers issues in the production line that are initially dismissed by the plant engineers as implausible based on their knowledge of engineering principles, only to be later confirmed to have been correct due to secondary and non-linear effects that their physical models did not take into consideration.

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

A:  It is much easier to cook a good meal if you have good ingredients. Knowing what data to collect and how to collect it makes all the difference in the world. However, even with incomplete and noisy data, it is possible to extract useful insights, as long as you account for the limitations of the inferences that can be deduced from such a data. With a tough piece of meat you can’t grill a nice steak, but you can still make a great stew.

————————

Don't miss Carlos’ conference presentation, Manufacturing Analytics at Scale:  Data Mining and Machine Learning inside Bosch, at PAW Manufacturing, on Wednesday, June 22, 2016, from 4:20 to 5:05 pm. Click here to register for attendance. 

By: Bala Deshpande, Founder, Simafore and Conference Co-Chair of Predictive Analytics World for Manufacturing.

No Comments yet »

March 25th 2016

Wise Practitioner – Predictive Workforce Analytics Interview Series: Pasha Roberts at Talent Analytics, Corp.

By: Greta Roberts, Conference Chair, Predictive Analytics World for Workforce 2016

In anticipation of his upcoming Predictive Analytics World for Workforce conference presentation, Predictive Job Maps – Modeling an Entire Workforce with a Tree of Models, we interviewed Pasha Roberts, Co-Founder and Chief Scientist at Talent Analytics, Corp. View Pasha_Roberts imagethe Q-and-A below to see how Pasha Roberts has incorporated predictive analytics into the workforce of Talent Analytics, Corp. Also, glimpse what’s in store for the new PAW Workforce conference.

Q: How is a specific line of business / business unit using your predictive decisions? How is your product deployed into operations?

A: Our clients use our predictions for performance or attrition every day, in the process of recruiting new job candidates. At PAW, I will be talking about rolling these up into a full network or tree of predictions, to predict performance for the candidate’s likely future roles, after a promotion or two. For this, we’ve built several dynamic job maps, and several composite benchmarks, and will soon deploy the full integrated solution that rolls all of this together.

Q: If HR were 100% ready and the data were available, what would your boldest data science creations do?

A: I’d create a living, interactive, visual map that shows everyone at the company who is going where, and how it affects corporate value. I’d populate the map with hundreds of predictive models to optimize employee changes, and keep it fresh with ongoing research. The result would be a living guide with tangible directives for hiring, terminating, and promoting employees to grow the organization right.

Q: When do you think businesses will be ready for “black box” workforce predictive methods, such as Random Forests or Neural Networks?

A: Not for many years, if they are kept opaque. HR and Hiring Managers want to know what is driving selection, and that it makes sense with a management narrative that they can follow. Sometimes there is a tug-of-war between plausibility and accuracy, but fortunately the human ability to form narrative is strong.

This doesn’t mean you are limited to simple regressions. Variable selection is everything for regression models, and we often use random forests or lasso/elastic net methods to find a set of regression variables that robustly predict.

Also, you can use black-box models like Random Forests, Support Vector Machines or Neural Nets for better accuracy as long as you do the extra work to isolate key variables and trends in them, to build a story for your model users. There are methods to probe a winning black box method to identify variable importance and general directionality. This can take enough opacity off of a model to be able to give “face validity” to its users.

Q: Do you have suggestions for data scientists trying to explain the complexity of their work, to those solving workforce challenges?

A: We often use examples from other domains, preferably their own domain. If I’m explaining a hiring model to a banker, we’ll show how it’s like using a credit score before offering a loan to someone. You don’t extend the loan (or job offer) unless there is a good probability that the person will pay the loan back (or stay on the job, or perform on the job). They get that.

Some of the technical graphs, like survival curves, work well with management users, and you can explain them; to most. Other constructs, like an AUC curve or cluster silhouette plots, are just not going to work with most. We try to win a lot of trust by the time we get to that stage.

Q: What is one specific way in which predictive analytics actively is driving decisions?

A: As mentioned above, hiring decisions – choosing candidates who are likely to not terminate early, and who are likely to overachieve (or not underachieve) real business KPIs. We work with clients to identify the KPIs – they are tangible things, like sales per month, or food safety scores, or cash drawer entries. We don’t try to drive mushy middle values like engagement or happiness.

You don’t put graphs or anything fancy in front of a recruiter for hiring/promotion decision. We just deliver calibrated, color-coded color bands – blue candidates are likely to over-perform, red candidates are likely to under-perform. They just pick up that color from the system; maybe get some auto-generated behavioral interview questions and talking points tailored to the candidates, and move to the next step in recruiting.

Real analytics ultimately doesn’t show graphs or paragraphs of “insights” to the user – it just helps them make the decision.

Q: How does business culture, including HR, need to evolve to accept the full promise of predictive workforce?

A: A predictive analyst needs to earn their trust, and they need to learn to understand that their “gut” is just another form of decision-making. The HR users need to gently see that sometimes the “gut” can be more biased, and less fact-driven than rigorous analytics.

They don’t need to learn Chi-Square or Receiver Operator Characteristic – though we’ve converted several into junior analysts and gotten a few back into graduate school. Managers do need to understand that models are just trying to make decisions based on facts, the way they are, and that they need to be forever learning.

Q: Do you have specific business results you can report?

A: In one case we reduced annual attrition from 84% to 48% in 5 months. That saves the banking client over $1 million a year in replacement cost and employee lifetime value.

In another case we increased the ability to hire successful candidates (as measured by passing a Series 7 exam) by 12% – that high-volume situation saved the client over $4 million a year in replacement cost alone.

———————————-

​Don't miss Pasha’s conference presentation, Predictive Job Maps – Modeling an Entire Workforce with a Tree of Models, at PAW Workforce, on Tuesday, April 5, 2016 at 10:25 to 10:45 am. Click here to register for attendance.  

By: Greta Roberts, CEO, Talent Analytics, Corp. @gretaroberts and Conference Chair of Predictive Analytics World for Workforce

 

1 Comment »

March 7th 2016

Dr. Data’s Music Video: The Predictive Analytics Rap

By: Eric Siegel, Founder, Predictive Analytics World

With today’s release of “Predict This!” – the rap video by yours truly (a former university professor trying to be a pop star) – I took the opportunity to ask a few questions of the music video’s featured character, Dr. Data.

Here is the video, followed by my interview with Dr. Data:

 

You can also access and share this video on:

 

facebook image

youtube

vimeo

 

Info, MP3 download, lyrics & more: PredictThis.org

 

INTERVIEW WITH DR. DATA:

Eric Siegel: Please summarize your vision for this rap video.

Dr. Data: Inspired by true events, this video recounts my origin story. More than a dance video for a song about predictive analytics, this film allegorically reconfirms the power of embracing your true inner self.

Eric Siegel: A pop song about predictive analytics? Why?

Dr. Data: Enlightenment. Infotainment tastes great and it’s good for you, too. A song can make a topic some find dry or intimidating fun and accessible. The more sophisticated viewer who listens carefully to the lyrics (click for full lyrics) will learn about this critical technology.

My goal was to think outside the quadrilateral parallelogram and make the best educational predictive analytics geek rap music video ever.

Done deal, due to a certain lack of competition.

And didn’t you yourself used to be a singing computer science professor?

Eric Siegel: Yeah, in the late 90s at Columbia University I would show up for a lecture with my keyboard and perform, for example, a rock ballad about the angst of debugging (online recordings are available). And in 1999, I published and presented on “Why Do Fools Fall Into Infinite Loops: Singing To Your Computer Science Class” (PDF of paper) at an education conference in Poland.

Dr. Data: Did your students like the songs?

Eric Siegel: Yeah, college students would much rather be at a rock concert, even a really bad one. One metric used in the above education paper was duration of applause.

Dr. Data: You, me, and Lady Gaga, we love that applause.

Eric Siegel: How many countries did you travel to for this video shoot?

Dr. Data: Well, truthfully, I was going to those countries anyway… but the video includes 10 locations across 5 continents – 6 countries plus Antarctica (did you notice the penguins?), which is a continent without any countries. Nor time zones or currency for that matter.

Eric Siegel: So, no green screen?

Dr. Data: Only for the outer space shots.

Eric Siegel: I noticed you play chess against a robot dancer in the video.

Dr. Data: More rap artists ought to also play chess with their dancers. In fact, that dancer (Claudine Quadrat) is actually a chess champ in real life, with a 1998 trophy from Kimball Wiles Elementary School to prove it.

Eric Siegel: That’s funny, cause I also was a childhood winner (at age 13, the 1982 Burlington, Vermont city chess champion of the “Booster” section, the lower of two sections across all ages).

Dr. Data: I’m glad things worked out in the end, what with the audience joining your party and your appearance on the cover of People Magazine.

Eric Siegel: It seems kind of absurd that, as Thomas Davenport and DJ Patil famously put it, data scientist is the sexiest job of the twenty-first century. Isn’t that status reserved for firefighters? But you, Dr. Data, really gave me courage with your mad flow and dance moves. I’ve always felt I was a pop star stuck in a geek’s body. And now I’ve got the moves like Jagger… errr… like his sound engineer.

Dr. Data: By the way, watch for the talented actor Nic Frantela in the video. He narrated the audiobook for Predictive Analytics: The Power to Predict Who Will Click, Buy, Lie, or Die – and can be seen as the first of two individual men shown to transform magically into geeks (and elsewhere in the video).

Also check carefully the People Magazine cover at the end for two true-life predictive analytics gurus who frequently speak at Predictive Analytics World: John Elder and Dean Abbott.

Eric Siegel: Who are you, Dr. Data? Where do you come from?

Dr. Data: Isn’t it obvious? I’m your alter ego, dude.

Eric Siegel: You mean I’ve been interviewing myself this whole time?

Dr. Data: Gimme a hug. It’s time you embrace your inner geek.

Eric Image 2015 croppedEric Siegel, Ph.D. (aka Dr. Data) is the founder of the Predictive Analytics World conference serieswhich includes events for business, government, healthcare, workforce, manufacturing, and financial servicesthe author of Predictive Analytics: The Power to Predict Who Will Click, Buy, Lie, or Die—Revised and Updated Edition (Wiley, January 2016), executive editor of The Predictive Analytics Times, and a former computer science professor at Columbia University. For more information about predictive analytics, see the Predictive Analytics Guide and follow him at @predictanalytic. Inquiries: eric@predictionimpact.com.

1 Comment »

« Prev - Next »

  • Subscribe Via Email

    Get a daily digest of new posts delivered to your inbox:

  • Predictive Analytics: The Power to Predict Who Will Click, Buy, Lie, or Die
    BOOK AWARDS:
  • Recent Posts