Monthly Archives: June 2017

Breaking Up the Boys’ Club to Unlock the Tech Industry’s Untapped Potential

June 29, 2017

teradata-analytics-humans1

A little while ago, our family car needed to go to the shop for repairs. I asked my wife if she would prefer to take it, but she refused. She worried that if the undoubtedly male mechanics saw her, they wouldn’t treat her fairly.

She’s not alone in imagining this is the case. In fact, there is now a female-owned auto shop in Pennsylvania that caters to women, offering salon services while they wait for an oil change. Patrice Banks, the owner and a former materials engineer, also hosts a monthly car clinic for women.

Banks’ business proves that all it takes is a concerted effort to break down diversity barriers. Her business has many parallels to science, technology, engineering and math fields in Silicon Valley. The technology industry has garnered a poor reputation for its diversity. When it comes down to it, auto repair is a technology sector. Its skills are housed in the same STEM umbrella as jobs in coding, big data and basically every other scientific discipline. And the social barriers that create an unconscious — and sometimes shamefully conscious — message that it is not friendly to newcomers has its roots in the same reason my wife doesn’t believe she has a fair shot at an auto shop.

One of Patrice Banks’ automotive class students put it simply: “I did not know that girls could go into the automotive field.” Technology companies can stop enabling this cultural message by making a real effort to be more diverse themselves.

As employees and creative thinkers in STEM fields, we pride ourselves on logical thinking. We imagine we are always hiring the best employee based entirely on facts, rather than on personal bias, and that this will result in the best workforce. But any good, logical thinker knows that you have to address your own biases to get real results. Science itself has proven that STEM is full of subtle biases that challenge our ability to be diverse.

To cultivate a successful environment for diversity to thrive, it’s not enough to ask those underrepresented to “lean in” and forge their own path ahead as an outlier. All businesses should encourage grassroots diversity in STEM, where we can foster a future workforce that better reflects society as a whole. In my experience, the long-term outcome is clear — diversity in STEM education leads to diversity in the workplace, making technology companies function better.

When successful companies set out to hire or staff a project, diversity needs to be at the forefront of their thinking. Personally, I have witnessed an environment where people produce better work by making a concerted effort to be inclusive. More players at the table from varied backgrounds ensures your products and solutions are more well rounded, and it avoids your company from falling prey to any of the embarrassing tech-industry missteps rooted in a homogenous workforce. Creating a wider network of employees that represents different genders, races, physical abilities, and economic and geographic backgrounds means that your products and services will mirror that diversity.

Technology is so ubiquitous that it’s shameful to think of it as a boys’ club, or any other demographic modifier. The people we imagine as our generation’s great modern thinkers are engineers and scientists. Being exclusive about who gets to contribute to future advancements only limits the way we can improve modern life. On a daily basis, each and every person on the planet is interacting with, benefiting from and demanding more from their technology. Not listening to those voices and failing to serve those needs is not just bad corporate citizenship, it’s bad business.


Oliver_RatzesbergerOliver Ratzesberger is executive vice president and chief product officer for Teradata, reporting to Vic Lund. Oliver leads Teradata’s world-class research and development organization and provides strategic direction for all research and development related to Teradata database, integrated data warehousing, big data analytics, and associated solutions.

Prior to this appointment, Oliver led the software teams for Teradata Labs, including the Teradata Database, Aster, Client tools, and Viewpoint as well as Hadoop integration. Joining Teradata in early 2013, Oliver has a powerful and impressive background in leading big data analytics initiatives and technology transformation for Fortune 500 organizations. For example, Oliver drove a large analytics effort to consolidate systems into a newly redesigned Unified Data Architecture. He has a successful background in driving analytics culture and advancing the value of high-volume and high-velocity analytics solutions. Oliver spent seven years at eBay, where he was responsible for its data warehouse and big data platforms. During his tenure at eBay, he led eBay’s expansion of analytics and was responsible for the co-development of the Extreme Data Appliance as part of eBay’s Singularity project, leading Hadoop engineering teams and driving the integration of Teradata and Hadoop.

Prior to eBay, Oliver worked for startups in software development and IT. In addition he has an extensive background in Open Source software development and has been an active contributor and committer to various projects.

In 1996, Oliver joined NCR Corporation as a senior consultant for data warehousing in Vienna, Austria – and in 1999 moved to Orange County, CA where he led professional service engagements for Telecom and eBusiness. As director of professional services, he was responsible for projects at Verizon, ATT, T-Mobile, eBay, and Excite, among others. Oliver earned his engineering degree in Electronics and Telecommunications from HTL Steyr in Austria. He lives in San Diego with his wife and two daughters.

Deep Learning for Executives: What Exactly is it Again?

June 28, 2017

teradata-devops

From Uber using past trips to predict its customers’ future habits to Facebook automatically tagging a picture you upload of your family, data is everywhere these days, and smart companies are using it to inform a better experience for their customers. Could the same be true for your company?

When it comes to making sense of big data, enterprises initially invested in machine learning. Simply put, machine learning uses algorithms to find patterns in data fed to it by humans. (There are resources out there for executives that want a high-level overview of this approach.) Typically, machine learning deals with data that is simpler. This low-dimensional data can be analyzed in light of a handful of factors. But eventually, some companies started amassing so much highly complex data — things like images, which are complex, since ordering their bits of data and dimensions of primary colors are important — that it was time for something more sophisticated — enter deep learning.

“Wait,” you may be thinking. “I’ve been using those terms interchangeably. Aren’t they the same thing?” That is a pretty widespread assumption, but deep learning is exactly as it sounds — deeper.

Deep learning uses a layered approach to make better decisions by constantly curating the data is it fed. Think of it like this: Machine learning is like when you would cram for a test in college by re-reading your notes. Deep learning is when a child is continually presented with the letters of the alphabet and slowly learns the trillions of ways to sequence those letters into words. In one example, previously identified data is interpreted. In the second, the interpreter realizes the potential of the data they are given.

Why is this important for a business? It can give your business data insights at a broader scope and higher level of fidelity for more complex use cases.

Traditional analytics only takes a business so far. But as companies amass more and more data, past algorithmic methods, like decision trees and linear regressions, may not up for the task, depending, again, on the use case. Think of it like trying to predict what color the next car driving down a street will be by keeping tabs on the ones that pass by before it. Decision trees could do so on a two-lane road with moderate traffic, but deep learning can do the same on a six-lane highway with cars going Autobahn speeds.

As deep learning progresses, its sophistication is getting more and more human-like. By teaching machines to optimize input based on reward — the AI equivalent of giving a dog a treat for sitting — Google trained up its program DeepMind to beat humans at the game Go. This was once an unthinkable bar for deep learning to pass, with the 19-by-19 square board offering much more complexity than the computer-human chess matches from the IBM Deep Blue days.  

While deep learning progresses to higher and higher accuracies, it is getting really good at things like identifying symbols, words and letters. Take a spin through some of Google’s A.I. Experiments, like the often hilarious “Quick, Draw!” and see for yourself. The same technology that lets Google know you just drew a horse lets its self-driving cars know what a cyclist looks like. And while deep learning is still lagging in areas like natural language recognition, all it’s going to take is more practice for the machine to get it right.

So how do you know if deep learning is right for your business? For starters, you need a lot of data for it to work, otherwise it’s a lot like taking a rocketship to the grocery store. It’s a powerful tool, but you need a really complex problem to effectively use it.

So, is deep learning necessary for your business? Can it help drive revenue and profitability? Come back next time where I’ll review why adopting deep learning is important for your business.

For more on deep learning, check out this blog post.


MoPatel_Headshot_ResizedMo Patel is Practice Director of AI and Deep Learning at Teradata. In his role, Mr. Patel is focused on building the Artificial Intelligence & Deep Learning consulting practice via mentoring and advising Teradata clients and providing guidance on ongoing Deep Learning projects. Mo Patel has successfully managed and executed Data Science projects with clients across several industries, notably Major Cable Company, Major Auto Manufacturer, Major Medical Devices Manufacturer, Leading Technology Firm and Major Car Insurance Provider. A continuous learner, Mr. Patel conducts research on applications of Deep Learning, Reinforcement Learning, and Graph Analytics towards solving existing and novel business problems.

Deep Learning: New Kid on the Supervised Machine Learning Block

June 27, 2017

machine learning

In the second instalment of this blog, we introduced machine learning as a subfield of artificial intelligence (AI) that is concerned with methods and algorithms that allow machines to improve themselves and to learn from the past. Machine learning is often concerned with making so-called “supervised predictions,” or learning from a training set of historical data where objects or outcomes are known and are labelled. Once trained, our machine or “intelligent agent” is enabled to differentiate between, say, a cat and a mat.

The currently much-hyped “deep learning” is shorthand for the application of many-layered artificial neural networks (ANNs) to machine learning. An ANN is a computational model inspired by the way the human brain works. Think of each neuron in the network as a simple calculator connected to several inputs. The neuron takes these inputs and applies a different “weight” to each input before summing them to produce an output.

If you’ve followed this so far, you might be wondering what all the fuss is about. What we have just described — take a series of inputs, multiply them by a series of coefficients and then perform a summation — sounds a lot like boring, old linear regression. In fact, the perceptron algorithm — one of the very first ANNs constructed — was invented way back in 1957 at Cornell to support image processing and classification (class = “cat” or class = “mat”?). It was also much-hyped, until it was proven that perceptrons could not be trained to recognize many classes of patterns.

Research into ANNs largely stagnated until the mid-’80s, when multilayered neural networks were constructed. In a multilayered ANN, the neurons are organized in layers. The output from the neurons in each layer passes through an activation function — a fancy term for an often nonlinear function that normalizes the output to a number between 0 and 1 — before becoming an input to a neuron in the next layer, and so on, and so on. With the addition of “back propagation” (think feedback loops), these new multilayer ANNs were used as one of several approaches to supervised machine learning through the early ’90s. But they didn’t scale to solve larger problems, so couldn’t break into the mainstream at that time.

The breakthrough came in 2006 when Geoff Hinton, a University of Toronto computer science professor, and his Ph.D. student Ruslan Salakhutdinov, published two papers that demonstrated how very large neural networks could work much faster than before. These new ANNs featured many more layers of computation — and thus the term “deep learning” was born. When researchers started to apply these techniques to huge data sets of speech and image data — and used powerful graphics processing units (GPUs) originally built for video gaming to run the ANN computations — these systems began beating “traditional” machine learning algorithms and could be applied to problems that hadn’t been solved by other machine learning algorithms before.

Neural Networks

Milestones in the development of neural networks (Andrew L. Beam, http://beamandrew.github.io/deeplearning/2017/02/23/deep_learning_101_part1.html)

But why is deep learning so powerful, especially in complex areas like speech and image recognition?

The magic of many-layered ANNs when compared with their “shallow” forebears is that they are able to learn from (relatively) raw and very abstract data, like images of handwriting. Modern ANNs can feature hundreds of layers and are able to learn the weights that should be applied to different inputs at different layers of the network, so that they are effectively able to choose for themselves the “features” that matter in the data and in the intermediate representations of that data in the “hidden” layers.

By contrast, the early ANNs were usually trained on handmade features, with feature extraction representing a separate and time-consuming part of the analysis that required significant expertise and intuition. If that sounds (a) familiar and (b) like a big deal, that’s because it is; when using “traditional” machine learning techniques, data scientists typically spend up to 80 percent of their time cleaning the data, transforming them into an appropriate representation and selecting the features that matter. Only the remaining 20 percent of their time is spent delivering the real value: building, testing and evaluating models.

So, should we all now run around and look for nails for our shiny, new supervised machine learning hammer? We think the answer is yes — but also, no.

There is no question that deep learning is the hot new kid on the machine learning block. Deep learning methods are a brilliant solution for a whole class of problems. But as we have pointed out in earlier instalments of this series, we should always start any new analytic endeavour by attempting to thoroughly understand the business problem we are trying to solve. Every analytic method and technique is associated with different strengths, weaknesses and trade-offs that render it more-or-less appropriate, depending on the use-case and the constraints.

Deep learning’s strengths are predictive accuracy and the ability to short-circuit the arduous business of data preparation and feature engineering. But, like all predictive modeling techniques, it has an Achilles’ heel of its own. Deep Learning models are extremely difficult to interpret, so that the prediction or classification that the model makes has to be taken on trust. Deep learning has a tendency to “overfit” — i.e., to memorise the training data, rather than to produce a model that generalizes well — especially where the training data set is relatively small. And whilst the term “deep learning” sounds like it refers to a single technique, in reality it refers to a family of methods — and choosing the right method and the right network topology are critical to creating a good model for a particular use-case and a particular domain.

All of these issues are the subject of active current research — and so the trade-offs associated with deep learning may change. For now, at least, there is no one modeling technique to rule them all — and the principle of Occam’s razor should always be applied to analytics and machine learning. And that is a theme that we will return to later in this series.

For more on this topic, check out this blog about the business impact of machine learning.


Martin_WilcoxMartin Willcox – Senior Director, Go to Market Organisation (Teradata)

Martin is a Senior Director in Teradata’s Go-To Market organisation, charged with articulating to prospective customers, analysts and media organisations Teradata’s strategy and the nature, value and differentiation of Teradata technology and solution offerings.
Martin has 21 years of experience in the IT industry and is listed in dataIQ’s “Big Data 100” as one of the most influential people in UK data-driven business. He has worked for 5 organisations and was formerly the Data Warehouse Manager at Co-operative Retail in the UK and later the Senior Data Architect at Co‑operative Group.

Since joining Teradata, Martin has worked in Solution Architecture, Enterprise Architecture, Demand Generation, Technology Marketing and Management roles. Prior to taking-up his current appointment, Martin led Teradata’s International Big Data CoE – a team of Data Scientists, Technology and Architecture Consultants tasked withassisting Teradata customers throughout Europe, the Middle East, Africa and Asia to realise value from their Big Data assets.

Martin is a former Teradata customer who understands the Analytics landscape and marketplace from the twin perspectives of an end-user organisation and a technology vendor. His Strata (UK) 2016 keynote can be found at: https://www.oreilly.com/ideas/the-internet-of-things-its-the-sensor-data-stupid and a selection of his Teradata Voice Forbes blogs can be found online, including this piece on the importance – and the limitations – of visualisation.

Martin holds a BSc (Hons) in Physics and Astronomy from the University of Sheffield and a Postgraduate Certificate in Computing for Commerce and Industry from the Open University. He is married with three children and is a lapsed supporter of Sheffield Wednesday Football Club.  In his spare time, Martin enjoys playing with technology,flying gliders, photography and listening to guitar music.


Frank SauberlichDr. Frank Säuberlich – Director Data Science & Data Innovation, Teradata GmbH

Dr. Frank Säuberlich leads the Data Science & Data Innovation unit of Teradata Germany. It is part of his repsonsibilities to make the latest market and technology developments available to Teradata customers. Currently, his main focus is on topics such as predictive analytics, machine learning and artificial intelligence.

Following his studies of business mathematics, Frank Säuberlich worked as a research assistant at the Institute for Decision Theory and Corporate Research at the University of Karlsruhe (TH), where he was already dealing with data mining questions.

His professional career included the positions of a senior technical consultant at SAS Germany and of a regional manager customer analytics at Urban Science International.

Frank Säuberlich has been with Teradata since 2012. He began as an expert in advanced analytics and data science in the International Data Science team. Later on, he became Director Data Science (International).

Why The CFO Cannot See The Value Of Data And Analytics In The Balance Sheet

June 26, 2017

value

What’s it worth?

During a recent business-value consulting engagement with a global telecom company, the CEO told me he believed that the investments he was making in enterprise information assets were generating huge value. The only problem was he could see no evidence of it other than the retained revenue and revenue growth reported by his marketing team. Shortly after, the director of BI and analytics confided that, in his opinion, the advanced analytics team was a lot more valuable than the company’s information assets, and wanted to know if there a way to quantify the team’s worth.

And then, on an entirely different occasion, the CCO of a multi-national, mobile-service provider asked if it was possible to quantify the value of their data and analytics.

For both of the above engagements, business-value audits demonstrated data and analytics ROI in excess of IRR 330 percent; value which was not reflected in the enterprise performance reporting or the balance sheet. Surprisingly, many commercial and non-profit enterprises find themselves in the same boat.

So, why focus on value?

Data, BI, and analytics, have always been seen as necessary infrastructure and foundational investments. Therefore, they’ve been regarded as part of the cost of doing business – in other words, overheads. And while classified as overheads, the very investments that could help an organisation keep its head above water are at most risk from being cut.

Now, companies are realising that data and information provide a ‘fingerprint’ for customer product usage. They understand that customer interaction can be tracked and analysed across all channels, a benefit which represents a significant competitive advantage for the enterprise. And this realisation has shifted the economic emphasis onto ‘value’.

That said, it’s difficult for organisations to measure the value of data and analytics due to a wide range of factors, including shortfalls in generally-accepted accounting principles and practices.

How can analytic value just vanish into thin air?

The financial benefits resulting from data and analytics can be quantified as revenue growth, cost reduction, margin improvement, cost avoidance, revenue recovery, and cash flow improvement. These classifications eventually show-up in the bottom-line, appearing in the enterprise balance sheet as general reserves, and/or the net profit & loss account. In essence, the assets minus liabilities in the balance sheet are balanced by shareholder funds, general reserves, and net profit & loss.

However, the value generated by data and analytics is neither directly attributed to the top- nor the middle-line and, therefore, sinks to the bottom-line and disappears without trace.

The problem of value attribution

Consider the two examples below and see for yourself where the value of data and analytics went and, within your own organisation, to whom the value would be attributed.

In a deregulated energy and utilities market, the price of the highest bid required to meet demand, dictates the 30-minute clearing price for the wholesale energy market. Energy retailers have recognised that, as volatile spot prices change every 30 minutes, quarterly meter readings do not allow them to forecast usage and demand, potentially creating an unprofitable outcome. Fortunately, granular Interval Meter usage data together with corresponding weather data enables them to develop accurate predictive modelling. This helps them stay competitive and profitable in a free market.

Here’s another example from other industries where although, publically, companies don’t admit they’re losing revenue due, internal audits show that fraudulent activities are costing them up to eight percent of their annual revenue. For an organisation with a billion dollar annual revenue the loss is at least $30 million. Even if only ten percent of this is recovered, that’s at least $3 million top- and bottom-line improvement after a comparatively small investment in analytics (applied to the data the organisation already owns) of, say, $100,000.

Anti-money laundering, SIM Box, and IDD Premium Rate numbers, are known areas of fraudulent activity that require advanced analytics to detect ever-evolving patterns and allow preventive action to be taken in time.

Are data and analytics corporate assets?

The treatment of data as a balance sheet asset is also wrought with difficulties. Although data and information continue to provide growing current and future economic value to the enterprise, they are never captured as either tangible or intangible assets. Instead, the hardware and software that make-up the bulk of the initial investment end-up as tangible assets, with a book value that bears purchase cost and is subject to depreciation. Data scientists and analysts are never treated as assets. They appear as costs in the HR system that roll into profit & loss statements.

Data and analytics projects are either capitalised – appearing as tangible assets and depreciated – or written-off as expenses, destined to appear in the profit & loss account.

Crude oil, gold mine, or data lake

This notion of big data as a mineral or crude oil suggests that data is a natural resource waiting to be exploited. And that this natural resource has a known up-front value which reduces in value over time (depleting asset).

But data has no pre-set value. And I think data and analytics is an investable growth asset that propagates in volume and variety, in line with competitive pressures. Which means it can be enriched to differentiate the business and drive greater business value.

For me, big data is the cumulative effect of small increments of data thrown up during innovative responses to competition – whether that’s while winning a share of the customer wallet or competing for scarce funds in not-for-profit sectors. Critically, business users need to be able to get hold of, optimise, and analyse big data, which means configuring and monitoring data pipelines in and through the data lake so they have constant access to high-quality data.

Gaps in GAAP

Accounting principles (e.g. GAAP / IFRS) are no help either – not with value recognition nor the valuation of data and analytics (e.g. difficulties of assigning value of human talent in the books, and shortcomings of valuing enterprise data as an asset in either tangible or intangible assets). The simple fact of the matter is that the balance sheet provides no direct visibility of data and analytics assets.

Recognising the value of data and analytics

Now we’ve addressed why the value of data and analytics are not directly evident in an enterprise’s financial performance reporting, keep an eye out for my next blog, ‘How to make the value of data and analytics visible to the CFO’.

Part two of this series can be found here.


SundaraRamanSundara Raman, Senior Telecom Industry Consultant, Teradata Australia

Sundara has been a Telecom professional for the over 30 years with a wide range of interests and multi-national experience in product management, solution marketing, presales for new generation networks and services, information management strategy, business intelligence, analytics and enterprise architecture development.

At Teradata, Sundara focuses on Business Value Consulting, Business Intelligence, Discovery Analytics and Customer Experience Management solutions.

Sundara has a Master’s Degree in Business and Administration with research on economic value of information from Massey University, New Zealand.

For the last 20 years, Sundara has been living in Sydney, Australia. In his spare time, Sundara enjoys walking and maintaining an active life style. Sundara is an inventor and joint holder of an Australian patent with his clinical psychologist wife. The invention is an expert system in cognitive mental health that applies machine learning algorithms.

 

Open Banking – For Whom?

June 23, 2017

RS3875_shutterstock_588093767If I wanted to explain the benefits of open banking to my teenage son, where would I begin? With the opening hours of our local branch? Or the 24/7 access he has on his mobile banking app? Unlikely – I can just imagine his response.

No. I’d say something like “suppose our bank received information from its customer base that young people like you needed a particular financial product or service – a credit card that keeps track of the CO2 load of each purchase perhaps, or a new type of savings account that gives savings credits based on the time you spend on Snapchat.”

“And suppose the bank said they wished they were in a position to provide it, but the teenage-product cupboard was bare. Then the bank contacted another company – a fintech or insuretech – to ask if they could buy that product from them and then offer it to you.”

Of course, the bank would need to work out a practical way to adopt the new product or service. Years ago, this was called white labelling. Back then, the connection between a bank and its supplier was managed manually by the back office, with a flag in the product admin system saying that there was an external provider.

Today, the connection between buyer and supplier is conducted via APIs (Application Programming Interfaces); the buyers being the incumbent banks and the suppliers, third-party providers more often than not.

So, who are banks open to exactly?

To everyone – vendors, suppliers, third parties, other companies and businesses, customers – in many and various digital ways. In fact, anyone who can provide data, a product, or a service, which the incumbent bank doesn’t have to develop itself (because of banking strategy or being unable to fit it in alongside the backlog of other development projects – regulatory or not).

Banks operate in a number of different ways:

Bank Channel. BaaC.
The old traditional way – servicing customers through own-front applications (entrance doors) by using APIs or a file as a channel for own-brand products.

Banking with API Market. BaaS.
Exposing services and products to third parties via open APIs. Customers show up from the back-end using the bank’s product or service, without interacting with the bank itself.

Banking as a Distributor. BaaD.
Integrating or bundling external financial services with own-bank offerings.

Banking as a Platform. BaaP
Fidor Solutions (a fintech), a bank itself, has a platform that offers a core of products and services to banking start-ups. I believe it’s just a matter of time before major banks realise that they can make use of their core to create new revenue in this way.

Banking as a Data Aggregator. BaDA
In future, data will be a commodity. Customers will own their data and will be able to tell data intermediaries and aggregators “I want the mortgage and deposit account with you, but my data is with a data intermediary x”.

One of the most interesting GDPR concerns is around data portability. The commissioner has really put the customer in the driver’s seat. And in the end, the providers that customers are most likely to trust with their personal data will be traditional banks, because they can demonstrate that they’re already holding enormous amounts of customer data, safely. The financial newbies will have a hard time convincing customers that they are as secure, reliable, and trustworthy.

Banking with Customer Collaboration. BwCC
Instead of leaving customers to pick their own path through the myriad of API-connected multiple parties and services, banks should take ownership of the relationship. They need to customise and personalise all those enabled open-banking services; their own, as well as others’.

Where do the incumbent banks position themselves?

Most reach the Banking As A Distributor level and then stop, thinking that the next levels are reserved for fintechs. They couldn’t be more wrong. With a laser focus on data (their most precious asset) and the right data ecosystem, major banks could create 3600 flexibility in their business model without having to share revenue with the disruptors. The winners will be banks that can utilise their data, drive down costs, build effective partnerships, provide robust security and, most importantly, maintain the highest data ethics.

So, do my son and his friends care about open banking? Hardly. Open banking is just another way of operating the machinery and embracing emerging technologies like AI, IoT, and Blockchain.

In fact, most customers couldn’t care less about what happens under the hood as long as it makes their day-to-day financial journeys easier.


AnetteBAnette Bergendorff is Senior Business Consultant, NORDEE, at Think Big Analytics, a Teradata company. Prior to Teradata, Anette worked for SEB Group between the years of 2001 and 2017, first as a business developer and then as a business architect. Annette has 23 years’ experience working in Finance and Insurance, and her roles have covered a range of areas such as business development, international partnerships and sales. Annette sees that key challenges facing the financial industry, such as GDPR, require a better approach to customer trust and data ethics: two key differentiators in terms of driving competitive advantage.

Big Data Brings Recruitment into the 21st Century

June 22, 2017

recruitingUntil quite recently, recruitment methods and motivation were underdeveloped and carelessly implemented. A vague urge to stay clear of trouble and to make sure the company did the ‘right thing’ in terms of meeting targets, was inefficient and unproductive. Driven by emotion and historical best practice instead of using hard data and information, diversity initiatives often failed to identify the up and coming employees that companies want to work with.

Research shows that increased diversity within a company has a positive effect on business results and market share, not to mention sales. As elements of the business come to have more of a provable impact, they rightly become strategised as an imperative function. Analysing the recruitment process can help businesses identify who will be the high-fliers of the future, and make sure they are meeting the requirements of an increasingly diverse talent pool.

There is increasing diversity in society, and diverse subgroups are using ever-wider bases of social media. Being across those is vital for companies to accurately and measurably hit targets. As the definitions around diversity widen, methodologies have to tally this with the expectations of Generation Z now beginning to enter the workplace.

Companies using an evidence-based approach for monitoring their hiring funnel around vital demographic information (such as age, gender and ethnicity), are giving themselves a better chance of success. Those in charge of diversity recruitment can, and should, use data and advanced analytics to answer questions that were previously done ‘by feel’. Solid info can inform organisations on what the best employer branding approach is, or the most effective candidate closing approach. This means companies not only have an increased awareness of how to engage with diverse communities, but how to make sure their ideal worker ends up working for them!

Just a few of the areas companies should be collecting data on to optimise their recruitment drives are: how people hear about job roles, attraction and turnoff factors, preferred communication channels, even down to keywords used in a recruitment drive. Whilst other areas of a business move towards data-driven modelling, recruitment (especially when it comes to diversity), seems to be stuck in the Stone Age. A call from the c-suite for modernisation and a ‘coming into line’ with the rest of the businesses data-driven core strategies, is sorely needed.

If employers want diversity results they need to measure it, reward it, and use data to continually update their recruiting approaches. Unless methods can show their effectiveness with hard evidence in the form of data, they should be reduced and removed. Data shows companies what’s working, and by focusing on those methods, they can ensure optimised initiatives. Ultimately, using outdated 20th-century approaches and tools in a 21st-century recruitment marketplace will simply not work. To better achieve diversity, companies need to utilise data. It opens up new methods, refines existing processes, and makes for happier, and more profitable companies.


3DE979DC-C174-4638-AD6A-E033383CE6ABAlison McCaig is International HR Director at Teradata, a position she has held since 2012. Previously, Alison worked as EMEA HR Director at Teradata between 2007 and 2012. Prior to this, Alison was the European HR Director at Life Technologies.

Building the Machine Learning Infrastructure

June 21, 2017

constructionMaking intelligent and accurate predictions is the core objective of machine learning and artificial intelligence applications. To achieve that objective, the machine learning or artificial intelligence application needs clean and well-organized information in a robust ecosystem architecture.

Machine Learning (ML) is the process of a computer system making a prediction based on samples of past observations. There are various types of ML methods. One of the approaches is where the ML algorithm is trained using a labeled or unlabeled training data set to produce a model. New input data is introduced to the ML algorithm and it makes a prediction based on the model. The prediction is evaluated for accuracy and if the accuracy is acceptable, the ML algorithm is deployed. If the accuracy is not acceptable, the ML algorithm is trained again with an augmented training data set. This is just a very high-level example as there are many factors and other steps involved.

Machine Learning Example

Artificial intelligence (AI) takes machine learning to a more dynamic level producing a feedback loop in which an algorithm can learn from its experiences. In many cases an intelligent agent is used to perceive an environment and detect changes in the environment and then reacts to that change based on information and rules it has been taught.

Every AI program is dependent on information to make predictions and decisions. That information needs to be structured in the appropriate context to make informed decisions.

An example of appropriate context comes from an example application of a robotic vacuum cleaner [1] that would navigate a room on its own and how it was measured that it was doing a “good job”. The metric chosen was focused on “picking up the dirt” and therefore to measure the volume of dirt it vacuumed and the amount of time it spent collecting it. Based on this objective the vacuum would learn that when it bumped into an object dirt would get picked up, and thus it learned to identify where the most dirt was collected next to furniture or some other object and would bump the object harder to dislodge any additional dirt, such as knocking over a plant and dumping the dirt on the floor and then collecting it. It consumed more energy which in turn cost more, not to mention causing a mess, but it did a “good job” based on the metric by which it was measured. It based this on the context of the information to which it had access.

Keeping this type of approach continued to increase expenses and decrease benefits.

The solution was to change the perspective to a new metric of “clean the room and keep it clean” and thus the application learned to just focus on expending energy only in the areas that needed to be vacuumed and reduced the cost of energy consumed by the device. It needed additional sensors to accomplish this new mission which at first sight would seem to increase cost, but the reduction of energy used was paid back with each occurrence producing significant value. It functioned on the terms of efficiency.

For AI, machine learning, and any type of analytics, the better the information is modeled, structured and organized for fast retrieval, the more effective and efficient the processing will perform.

Conversely the more complex the model or structure, the more complex the processing.

334-Think-Big-Analytics-RGB-KYLO-Diagram-v12

AI and ML algorithms that search for patterns in unstructured or non-relational data still need structure. Even schema-less data must be wrangled into meaningful structures. AI and ML algorithms are most effective when the enterprise architecture enables efficient access and retrieval of information for specific contexts. The ingestion framework for an enterprise ecosystem architecture needs to consider the information and data needed for machine learning and analytics. The landed data should be a single usage point where data can be used across multiple applications and platforms, in other words land once, use many.

Kylo is an open source solution for data ingestion and data lake management employing NiFi templates to build an ingestion pipeline with cleansing, wrangling, and governance to transform data into meaningful structures needed for machine learning and analytics.

Kylo Workflow

Kylo provides an ingestion framework that is a key component of any machine learning infrastructure. It leverages Nifi and Spark and is flexible to add others. The ingestion framework includes a wrangling component that facilitates the transformation of data into meaningful structures that ML and AI will rely on to make enhanced predictions. Data lineage is also captured in the framework to enforce governance. The framework accelerates the development process and iterations critical in constantly improving model accuracy.

Boosting business outcomes with the best ML and AI applications truly relies on a robust machine learning infrastructure and a well-thought-out ecosystem architecture. Kylo is a Teradata sponsored open source project under the Apache 2.0 license that provides an extensible framework for the machine learning infrastructure. Teradata also provides an ecosystem architecture consulting service to harness the vast experience of technology professionals in combining the right mix of technologies and data platforms into an efficient digital ecosystem.

References

[1] S. Russell and P. Norvig, Artificial Intelligence: A Modern Approach, Upper Saddle River, New Jersey: Pearson Higher Education, 1995, pp. 46-61.

Pat Photo 2014

Pat Alvarado is a Teradata Certified Master providing technical consultation on analytic ecosystem architecture, workload distribution, and multi-genre analytics across multiple platform and analytics technologies.

Pat started his career as a hardware engineer building test instrumentation for mil-spec components and later point of sale systems for fine dining restaurants. After developing firmware for his micro controller hardware designs, Pat moved into software engineering developing data management applications with open source GNU software on distributed UNIX servers and disk-less workstations based on the Berkeley Software Distribution (BSD) as a departure from the proprietary AT&T UNIX and became known as FreeBSD.

Pat joined Teradata in 1989 providing technical education to hardware engineers on the DBC/1012 architecture and was part of the team building out the parallel software development environment on ClearCase.

Presently, Pat provides consulting and thought leadership on relational database management systems (RDBMS), Document and NOSQL database systems, Hadoop distributed file systems (HDFS), exploratory analytics platforms, etc. both on-premise and in the cloud via SQL, MapReduce, SQL-MR, Java, Python and other open source languages and architectures for structured, unstructured, and evolving schemas. Pat manages technical consultants in the development and implementation of data analytics and MapReduce extensions in Java, C++, R, etc. Development of relational and dimensional data models and universal modeling to bridge relational and non-relational schemas to support business processes.

Japan liaison for establishing relationships between U.S. and Japan organizations through business processes. Leveraging cultural approaches to manufacturing through continuous process flow and leveling workload (heijunka) and applying a continuous improvement process (kaizen).

The Outcome Economy, Powered by IoT

June 20, 2017

Missing piece

Twenty-five years ago, I was in business school when a professor gave us the assignment of forecasting the global demand for drill bits fifty years into the future.  My fellow students and I approached the problem in pretty much the same way, by making assumptions about how the world would be in that time, and what the impact would be on the drill bit marketplace.  We’ll be off fossil fuels by then, so no more need for those kinds of drill bits.  The population will be much larger, and that will drive demand for hand tools that rely on drill bits.  After everyone took their turns providing a number and the rationale behind it, the professor informed us that we were all wrong.  The answer, he explained, was that “fifty years from now, the world-wide demand for drill bits will be zero.  However, the world-wide demand for holes will be enormous!”

The point of this lesson was twofold.  First, that it is myopic to think that people need certain assets; rather what they need is the outcome of that asset.  People don’t need cars, they need mobility.  Cities don’t need street lights, they need streets that are safe to drive on and walk down at night.  People don’t need drill bits, they need holes.  Secondly, that this shift from buying products to buying outcomes would require emerging digital capabilities that we were just beginning to catch glimpses of 25 years ago.  These digital capabilities would enable companies to measure, analyze, and adjust their offerings in near real time in order to deliver and quantify their value.  Such outcomes may range from guaranteed machine uptimes on factory floors, to actual amounts of energy savings in commercial buildings, to guaranteed crop yields from a specific parcel of farmland.

Half way in, and we certainly appear to be well on our way to realizing that prophecy.  Enabled by increasingly rugged, low cost sensors, the physical world is becoming digitized.  Over the last 10 years, the digital exhaust from these sensor readings has enabled greater efficiencies, safety, and revenue opportunities.

Companies like Union Pacific were early beneficiaries by analyzing 20 million daily sensor readings that described the temperature and sounds from train wheel bearings.  Union Pacific can now predict a derailment with a high degree of confidence more than a week out, which has cut bearing-related derailments by 75 percent and reduced unscheduled maintenance-related delays.  Quite an achievement considering that a train derailment can cost upwards of $40M and put lives at risk.

Valmet has traditionally been a manufacturer of pulp grinding machines that produce tissues, glossy paper, cardboard and other paper products.  Valmet began instrumenting these machines – which are the size of a football field – to better understand what leads to unplanned downtime and inefficient consumption of machine consumables, such as belts, felts and chemicals.  The resulting data and analytics have led to two new revenue opportunities for Valmet.  First, they can deliver a service to clients on how to best optimize the machine for maintenance, which leads to higher uptime for their clients.  Second, they are able to quantify the value of their higher priced (and higher quality, as a matter of fact) consumables with respect to life expectancy under actual client operating conditions.

What we are starting to see now is that the industrial IoT leaders are establishing board level goals that go beyond operational efficiencies, safety, and add-on revenue streams to something much more disruptive and fundamentally game changing by selling outcomes.

Companies like Monsanto are moving from selling products like seeds and fertilizers, to precision agriculture where crop yields are maximized.  By connecting smart farm equipment such as tractors, tillers and seeders with data on weather, soil conditions, and crop health, Monsanto can measure, analyze, and adjust activities like when and how a farmer ploughs his field, how deep to plant the seed, and spacing of plants in a row.  Crops have their best chance to reach their highest potential when data from billions of events, coupled with combinations of analytic techniques involving statistics, machine learning, and graph analysis aid farm management practices.

Spanish train operator Renfe was looking to take market share from airlines on the route between Barcelona and Madrid.  Airlines at that time had 80% market share, due in most part to business travelers valuing the on-time performance of airlines compared to trains.  Enter Siemens, which didn’t just sell Renfe a train and a warranty; rather they continually monitor and resolve issues before they happen in order to deliver on the promise of reliable mobility.  A train developing abnormal patterns is dispatched for an inspection service to prevent failure on the track. This has resulted in only one out of 2,300 journeys being delayed by more than five minutes.  “That happens because we have data, we have analytics models, and we can actually predict certain failures,” said Gerhard Kress, Director of Mobility Services at Siemens.  “There’s gearboxes, for example, on high-speed trains, it’s one of the things that is most tricky to monitor. We had a couple of cases where you could predict those things would be breaking in a few weeks. We had ample time to provide the spare parts, do the right thing, repair it, take the train out of normal circulation without harming the schedule, and work with the customer without having any problems for them.”  Now, the airlines are down to 30% market share.  Siemens is increasingly selling more outcomes because they have the capabilities to measure, analyze, and adjust in order to deliver on the promise of that outcome.

Far from isolated case studies, we are seeing similar transformations based on outcomes at Boeing, Volvo, Maersk, and many others.   What they all have in common is they are industries that rely on heavy, complex assets where those assets are used by others to play a part in a much larger outcome.  That leads me to think that while my professor was prescient, Home Depot and Lowes will still sell simple hand held drills 25 years from now.


MeleyChad Meley is Vice President of Product and Solutions Marketing at Teradata, responsible for Teradata’s Big Data, IoT, and Machine Learning offerings.

Chad understands trends in the analytics, IoT, and big data space and leads a team of technology specialists who interpret the needs and expectations of customers while also working with engineers within Teradata Labs, consulting teams and technology partners.

Prior to joining Teradata, he led Electronic Arts’ Data Platform organization that supported Financial Analysis, Game Development Insights, and Marketing Analysis and CRM.  Chad has held a variety of other roles within data warehousing, business intelligence, Marketing Analysis, and CRM while at Dell and FedEx.

Chad holds a BA in economics from The University of Texas, an MBA from Texas Tech University, and performed post graduate work at The University of Texas.

Professional awards include Best Practice Award for Driving Business Results in Data Warehousing from The Data Warehouse Institute, Marketing Excellence Award from the Direct Marketing Association, and Marketing Gold Award from Marketing Sherpa.

Chad can be reached at chad.meley@teradata.com or on Twitter at @chad_meley

Working in the New World of Data and Analytics

June 19, 2017

RS3795_shutterstock_424934146Many enterprises are struggling with the complexity of today’s big data and data science ecosystem, though they recognize the opportunity of emerging practices. As a result, the shortage of trained data and analytics specialists remains one of the biggest challenges for organizations across the board. Mikael Bisgaard-Bohr, VP Business Development International, Teradata, participates in this Q&A to outline his perspective about working in the new world of data and analytics as well as what organisations need to do to target and recruit the right talent.

How did you land in your current job role?

I have been fortunate to work in what I call the ‘data industry’ for 20 odd years, and this is probably the most exciting time ever! It is more than 20 years since I finished my business degree and I remember on several occasions discussing work with fellow alumni in the years since.

While everyone else went to work for management consultants or investment banks, I was the only one who went into technology – and not a start-up, but what was perceived then as a stodgy ‘old’ technology company. Suffice it to say that I received significant encouragement to go and do something worthwhile and rewarding rather than ‘waste my time and talent on something as esoteric as data’. However, the tables are turning: what was once seen as boring is becoming relevant. Over the last 5-10 years several of my former class mates have called me and started asking questions about this whole “data thing” and a few of them are now working in the data industry themselves.

What does your job involve on a day-to-day basis?

I am responsible for Teradata’s long-term direction and strategy for the region with a specific focus on identifying new markets and industries for Teradata to enter. I have exposure to leading companies globally, which gives me a unique understanding and perspective on how global leaders are leveraging data and analytics to better compete in their markets. I interact with the largest and most sophisticated users of technology to see how it is changing the way companies are run, products are consumed and the interactions between organisations and consumers.

What are the origins of big data – how did it come about?

According to most sources that I have seen it was Clive Humby, one of the brains behinds Tesco’s successful loyalty scheme who coined the phrase ‘data is the new oil’ 10-15 years ago. It is an interesting analogy which shows that he, as one of the early pioneers of data-driven marketing, fully understood the value of data to the enterprise. The idea that data can be seen as the new raw material for the digital enterprise of tomorrow has now taken hold, as we are seeing the increased instrumentation of everything and everyone…

What is the role of data and analytics in today’s digital era?

Today we are rapidly getting used to the fact that almost every object is being instrumented, and so becomes a platform for data generation which in turn can be used to drive new and innovative services, apps or even completely new business models. But what is interesting is that in all these discussions we still tend to focus most of our attention on the volume of data. Just focusing on the fact that we can now get data on anything or anyone – and that the data volumes are increasing – misses the real revolutionary aspect of what is going on.

We are seeing organisations actively hiring and recruiting for a new role – that of Chief Data Officer – what will this role bring to the enterprise of the future?

I believe that in a lot of organisations there is an amount of “fashionable awareness” of certain roles and this is certainly true when it comes to the CDO. This means that for many organisations the CDO role will have little influence and consequence.

However in organisations that put data at their core, the CDO role becomes a critical leadership role focusing on preparing the organisation to leverage data as a competitive advantage by:

  1. Building adequate and relevant data management capabilities across the organisation by focusing on acquiring and training talent for analytics, building the necessary organisation that makes these capabilities relevant for the business and organisational frameworks to ensure that best practices are captured and distributed
  2. Creating an infrastructure within the organisation to ensure that data can be leveraged for better results
  3. Fostering a data driven culture across the organisation to ensure that data and analytics are embedded into a wide variety of business processes and that testing new business ideas with data becomes the new norm (as opposed to relying on ‘gut based’ decision making)
  4. Aggressively importing ideas from outside as well as sharing best practices to foster innovation
  5. Instilling a culture of courage to act on the results

The fact that there is a skills shortage in technology globally and more recently in data and analytics is widely written about. What can employers do to bridge this skills gap?

Historically many organisations have relied on external trainers to provide skills upgrade and training programmes for their workforce. It is now abundantly clear that there is a talent shortage in data and analytics, and as that demand continues to grow I am convinced that more external providers will jump in to fill the void.

But I don’t believe that this will be enough – an outside provider can train the workforce on basic analytical capabilities, and that has some value. However I strongly believe that for the workforce to become truly analytically enabled, organisations need to complement this with internal training – leveraging their own data and focusing more on the practical uses of that data.

What would your advice be to students and young graduates seeking roles the data and analytics space in today’s enterprise?

This is a very interesting question indeed. There is a wide spread perception that analytic skills are quantitative and therefore young graduates should focus on the quantitative aspects of analytics – statistics and quantitative methods. I agree that these are very important fundamental skills; without them it is very hard to become a strong analytic-driven associate. But some of the best data scientists and analytical people that I have worked with have had a very, very strong creative side as well, and I strongly feel that you cannot become successful with data and analytics without strong creative skills, as well as the conviction and courage to follow them.

Some of the most successful analytics cases I have encountered were not only successful because they drove business value, but uncovered new and previously unknown insights using methods that were not intuitive for that particular problem. What made these cases outstanding was the creative use of analytics coupled with how they were presented, enabling a much more effective and impactful communication of the results. This is why I believe that the Teradata’s ‘Art of Analytics’ is such a great programme because it highlights the importance of the other side of the equation.

So my advice would be to pursue a combination of both quantitative and creative skills so that the students of today can generate ground-breaking results whilst effectively communicating them to stakeholders.


Mikael Bisgaard-Bohr is the executive vice president and chief business development officer, reporting to Vic Lund, president and CEO. Mikael is responsible for aligning resources – either direct or via alliance partners –to ensure that our field resources are well prepared to meet the needs of our customers. In addition, he is responsible for leading the global alliances and learning teams at Teradata. Mikael is responsible for identifying new trends and directions in the market of analytics, business intelligence, and big data. He interacts with the largest and most sophisticated users of Teradata technology as well as leading minds in the industry – resulting in a better understanding of how technology is changing the way companies are run, products consumed and the interactions between organisations and consumers. Mikael has been working with analytic technologies for the last 20 years; prior to his current role he was running Marketing and Business Development for Teradata’s International region. Before that he was a thought leading business consultant focused on the retail industry. Mikael holds an MBA from SDA Bocconi.

Team Effort Makes a Big Difference in the Community

June 16, 2017

RS695_TDCaresLogo

Teradata Cares creates opportunities for employees to engage with their local communities in three key areas:

*Improving education to help tomorrow’s technologists and business leaders understand the possibilities that technology provides
*Strengthening neighborhoods and communities
*Helping the environment and supporting corporate sustainability

Local Community Cares Champions can be found all over the world. These individuals plan and organize Teradata Cares programs in their local communities.

Each month, we will highlight some of the work being done by Teradata employees in communities worldwide.

TeamPictureThis month, we’re shining the spotlight on Teradata Pakistan. The members of TD Pakistan work alongside community volunteers and have a team focus on:

1 – Education
2 – Healthcare
3 – Unemployment

Under these programs they help individuals and deserving families to do group projects.  The efforts of the TD Pakistan team have now impacted more than 15,000 individuals.

Some of the recent projects include purchasing books, uniforms and shoes at start of new academic year.  Fund raising and sponsoring medical treatment of unprivileged individuals, considering the very limited support from government in healthcare for general public.  Providing respectable means for earning income such as purchasing auto-rickshaws for unemployed persons.

Other projects include: (view details of the below in their Newsletter)

  • Spending a day with kids in village schools mentoring, distributing gifts and talking to their parents for better rearing of their kids
  • Arrangement of blood donation camps, free medical camps, and helping blind people (in progress)
  • Spreading awareness for finding ways of self-employment. Sponsoring skill development in unemployed persons in vicinity

Team Pakistan’s Ongoing Partnership with the Kamahan Village Shool Project – Lahore, Pakistan

Talking to students in pre-nursery_1TD Cares Pakistan team is running their biggest project yet which is to bring street children into school in ‘Kamahan’ village near Lahore.  They currently have 65 students in the program and are going to enroll even more students.  This program began in March, 2016.

The team brought many of the students, whose ages range from 9 – 11 years, to school for the very first time.  This is indeed one the biggest achievements in this project.

This school is under close supervision by the TD Cares team and is happy to state that all students are being sponsored by Teradata employees, which is in fact a considerable fund raising activity in itself!

Furthermore, the TD Cares team collects detailed data regarding each student and follows a pre-defined screening process before sponsoring a student.  They also take monthly feedback from the school and meet all parents at least once a quarter to keep them motivated.  Almost all the students are from very poor families, with the fathers and mothers working as laborers such as waiters, rickshaw, house guards, or car washers.  Those are the lucky parents; some don’t have jobs at all.

The team would like to highlight that they have interviewed almost 100 students and their parents and are continuously working to reach and support more students.

Teradata Cares Pakistan Team – Congratulations and thanks to you all!

  • Bilal Bin Munir
  • Taimur Ahsan
  • Adnan Kazi
  • Badshah Rehman
  • Sheraz Arshad
  • Azhar Ali Siddiqui
  • Jibran Ahmed
  • Muhammad Bilal Anjum
  • Muhammad Aslam
  • Arslam Nadeem
  • Sameer Shahzad
  • Rabiah Parvaiz
  • Mahira Iqbal Malik
  • Rubina Khan
  • Mehreen Malik
  • Syed Naveed Shahzad
  • Imran KhanTeam photo - gudy bags preparation