Extraction of business value from your data depends critically on data quality.

We assign economic value to information commensurate with how much better we can do with it than we can without it. This summarizes the point of my previous discussion on Information Economics. Information economics represents the foundation of business analytics.

My simple demonstration in that installment featured an element of uncertainty. I stated in my introduction to data science that data science — and business analytics, by extension — can never completely remove uncertainty from business decision-making. Separating uncertainty into resolvable and non-resolvable components is the best it can do.

We can do things about resolvable components of uncertainty underlying our decision-making. We can resolve uncertainty by:

• Judiciously selecting the best information on which to base our decisions; and
• Maximizing the quality of the information actually used.

I demonstrated in my previous installment that more data is not always better. Bringing more data into our decision-making only helps if the new data gives us information that we do not already have.  I call attention here to the quality of data.

I begin this installment by offering a definition of data quality. I then demonstrate the assignment of economic value to quality.  I use an "elementary game" illustration along the lines of Information Economics.  I conclude with a foreshadowing of how data quality will influence subsequent installments in this series.

What do we mean by data "quality"?

Quality is one of those words that we use extensively without defining it. This blog's scientific approach requires that we define things precisely. I cannot formulate and scientifically test hypotheses about things that are not precisely defined.

An abstract thinker, I am fond of Robert Pirsig's¹ definition of quality. Pirsig fused the idea of esthetics — the study and characterization of beauty — with technical conformity. The esthetic aspect makes applying the scientific method more challenging.  (Pirsig might be challenged by my imposition of this constraint on my line of inquiry.)

Quality management occupies a prominent role in the Program Management profession. The Project Management Book of Knowledge (PMBOK)² contains extensive sections devoted to quality management. The PMBOK itself lacks an obvious, explicit definition of quality. 

A leading study guide for the Project Management Institute's (PMI) Project Management Professional (PMP) exam offers a passing definition: "Quality is defined as the degree to which the project fulfills requirements."³ This then leads us to the concept of quality as conformity, or compliance. This gets us closer to a useful definition of quality.

So, what constitutes quality in data? GS1 identifies five key attributes of data quality. GS1 is a non-profit organization dedicated to the design and implementation of global standards and solutions to improve the efficiency and visibility of supply and demand chains globally and across sectors. Their mission — focused on facilitating information charing — positions them well to speak with authority. 

What are these five attributes?  I summarize them here.  I also offer both business and technical examples.  I used slightly different terminology.  For ease of memory, these are the "Five Cs of data quality."

• Completeness.  A quality data set contains all of the expected information on which a business decision is based. A database administrator (DBA) would focus technically on whether all of the required fields in each record are populated.
• Conformity.⁴ A data set must, in its entirety, conform to a well defined semantic and logical framework. A DBA might talk about a "data schema", here. The schema captures the logical part. Business decision-making takes us beyond the DBA. Data derived from multiple sources must allow "apples versus apples" comparisons. We sometimes refer to this as "semantic" conformity.
• Correctness.  Data is used to represent facts, or states of nature. Correct data represents facts accurately. A DBA will focus on measurement or data entry errors. Correctness is usually an attribute derived from the data source. It is, consequently, difficult to define data systems capable of internally checking data correctness.
• Currency. We desire to make decisions based on data representing the actual state of nature. A designer of a data management system will assign version numbers and time stamps. Business decision makers may also concerned about whether observations are made frequently enough to capture rapidly changing circumstances.
• Confidence.⁵  Data confidence is related to "veracity", one of the key characteristics IBM assigns to "Big Data." A business decision maker considers whether the data source possesses the authority to make the assertion represented by the data. A designer of a data management system will handle this through seeking policies that designate data sources as "authoritative."

These five attributes — our "Five Cs of Data Quality" — provide us with a holistic framework for thinking about and managing data quality.  I now illustrate the Information Economics of data quality.

Illustration of how data quality affects the value of information.

Market analytics gives obvious examples to illustrate opportunity costs resulting from poor data quality. Let's extend the BMW dealership illustration used a couple of installments ago. I first used it to show the economic value of information. I'll show here how poor data quality dilutes that value.

Let's say that my market size is about 100,000 potential customers. I define my market as households with net income in excess of $85,000 within a specific geographic radius. Market research gives me the following features about households in my market:

• Household incomes, conforming to a power-law statistic distribution;⁶
• Age of oldest current automobile, exponentially distributed;⁷ and
• Automobile brand preferences, uniformly distributed.

Past history gives me a model I use to relate these features to the probability that I can sell them a new BMW. Figure 1 shows the results of the model.  It ranks households according to my estimate of the likelihood of sale.  It contains two curves, an "Actual", and an "Underestimated."  I'll turn to the "underestimated" curve shortly.

You see in Figure 1 that the top decile — ten percent — contains most of the effective market. This is the households that are really likely to buy from me. So, my effective market is actually only about 10,000 households. If the average sale generates $50,000, this market is valued at $500 million.

Figure 1 — Predictive analytics tells me which households are most likely to buy, based on income, age of current vehicle, and brand preference.

Let's say that I previously used targeted marketing to improve the sales likelihood by an average of ten percent for households in in the 65% to 85% range of probability of purchase. These households are the best candidates for a targeted marketing campaign. There are 1,669 households in this group. Targeted marketing increases the size effective market by approximately 1,660 households.

Let's now turn to data quality. Unbeknownst to me, about a quarter of my customer preference ratings are underestimated by 20%. This is a data correctness issue.  My perception of the market is, as a result, represented by the "Underestimated" curve in Figure 1.

What does this do to my business?  The curve in Figure 1 only appears to shift a small amount.  But the real effect is to reduce the size of my effective market is reduced by 230 households. This amounts to around $11.5 million in reduced revenue opportunity. 

Some of this loss arises from misidentifying the best potential customers for my targeted marketing. On one hand, I'm not applying targeted marketing to customers who might benefit from it the most.  On the other, I expend resource on others who are less likely to be swayed. 

Poor data quality leads to misdirection of my targeted marketing efforts. This is an information economics issue. The utility of my information — how much better I can do with it than without it — is directly related to data quality.

What, then, does data quality mean for enterprise business analytics?

We see here that quality influences the value of information. Data quality — characterized by the "Five Cs" — does not necessarily occur automatically. That quality is accompanied by cost is unsurprising. 

The PBMOK briefly addresses the cost of quality.⁸ These concepts extend to data quality management. Two data quality costs arise:

• Costs of conformance are accrued through expenditure of effort to explicitly address data quality objectives; and
• Costs of nonconformance arise from adverse results due to shortfalls in data quality.

My illustration above highlights an example of the second. Data quality nonconformance dilutes the economic value of information.

Enterprises should strive to balance these two costs. In other words, they should expend efforts towards data quality commensurate with the information value requirements of a specific business context. Information economics provides the basis for business case analysis about data quality.

Business critical information supports decisions about issues for which considerable value is at stake. Data for these applications demand high quality.  For example:

• Decisions about major investments often placing substantial amounts of capital at risk; and
• Financial reporting must be governed by financial controls regarding which legal requirements for attestation apply.⁹

An entire industry developed to provide specialized data management systems to address auditing and financial reporting requirements. These systems are Enterprise Resource Planning (ERP) systems. 

At the opposite end of the spectrum, inserting a single targeted advertisement into an individual user's web browsing session costs very little. Netflix movie recommendations or Amazon book recommendations each similarly cost very little. All targeted ad placements or consumer preference recommendations must collectively achieve an average accuracy. 

References.

¹ R. M. Pirsig, Zen and the art of motorcycle maintenance, New York:  Harper Collins, 1974, Kindle Edition, p. 204, http://goo.gl/si1ayP.

² Guide to the project management book of knowledge (PMBOK), fourth edition, Newtown Square, PA:  Project Management Institute, Chapter 8, http://goo.gl/anjxhd.

³ Rita Mucahy's™ PMP® Exam Prep, seventh edition, RMC Publications, 2011, p. 263, http://goo.gl/k0ml4D.

⁴ GS1 talks here about "consistency". I take the liberty combining two factors, consistency and "standards-based."

⁵ GS1 does not address data confidence.
⁶ N. N. Taleb, The black swan, New York: Random, http://goo.gl/Ermy9s. 
⁷ L. Kleinrock, Queuing systems, Volume I: Theory, New York: Wiley, 1975, pp. 65-71, http://goo.gl/Dz8mVE.

⁸ Guide to the project management book of knowledge (PMBOK), fourth edition, Newtown Square, PA:  Project Management Institute, §8.1.2, http://goo.gl/anjxhd.

⁹ e.g., "Internal control over financial reporting in exchange act periodic reports of non-accelerated filers," Security and Exchange Commission, 17 CFR Parts 210, 229, and 249, http://www.sec.gov/rules/final/2010/33-9142.pdf.

© The Quant's Prism, 2014

Analytics demonstration: How much information does your data contain?

Our current big data buzz is accompanied numerous points of excitement and controversy. The "more data" versus "better algorithms" is prominent among these. This debate sometimes resembles an old Miller Lite beer commercial. 

Multiple business analytics luminaries have weighed into the conversation.  These include:

• Omar Tawakol, CEO of Oracle's data marketing operation BlueKai, concludes in Wall Street Journal's AllThingsD that "If you have to choose, having more data does indeed trump a better algorithm;"
• Gartner's Andrew Frank emphasizes the nuance by asserting that "if you have to choose, maybe you’re on the wrong track;"
• Information Week's Larry Stofko leads the conversation away from the dichotomy, towards "first we must make small data seamless, simple, and inexpensive;" while
• Jeanne Ross, et al, assert in Harvard Business Review that the question is irrelevant: That "Until a company learns how to use data and analysis to support its operating decisions, it will not be in a position to benefit from big data."

This obviously is a very complicated issue.  It defies the "Tastes Great-Less Filling" dichotomy.  Very few initiatives provide simple, automatic Return on Investment (RoI). Big data analytics is no exception. 

More data only delivers marginal value if that data brings new information. I provide a technical demonstration in this installment. I try to show this in terms that don't require a mathematical PhD. 

I demonstrate this concept using a data set from the U.S. Bureau of Labor Statistics (BLS). This is the first of a series of quantitative demonstrations. The BLS data offers numerous opportunities for interesting data science demonstrations. I plan to spend much of the Autumn tinkering with it.

My data source:  Bureau of Labor Statistics' (BLS) O*Net.

O*Net is a career counseling service provided by the BLS. It is based on Holland's Occupational Themes. Psychologist John L. Holland theorized that our personality preferences dictate the types of occupations we are likely to enjoy the most. The Myers-Briggs Type Indicator (MBTI) and Herrmann Brain Dominance Instrument (HBDI) are based on similar theories.

O*Net users complete a survey of interests and lifestyle preferences.  The O*Net application applies analytics to identify occupations that are likely to be attractive to someone with specific sets interests and preferences.

O*Net's is based on a database that contains tables describing 923 different occupations. They are characterized in terms of:

• Interests (9 attributes);
• Knowledge (33 attributes);
• Abilities (52 attributes); and
• Skills (35 attributes).

Each attribute contains two numerical values: One for importance and the other for level. So, the model characterizes each of the 923 occupations in terms of 258 attributes — or features.

What makes this data set interesting?  Four factors attract me:

1. It is a well-structured data set, requiring limited preparation;
2. It's open, well-documented, and easily accessible;
3. Some degree of authoritative research underlies the data set; and
4. It seems to contain an awful lot of attributes by which to distinguish between 923 occupations.

This last factor is the point of the discussion here, today.  How many of these 258 features are really needed to describe 923 occupations.

I demonstrate below that most of the information* about the 923 occupations is contained in a small subset of the 258 features. I do not argue that 258 features is too many to guide job seekers or career changers. I make my point from an information theory point of view: More data attributes do not necessarily yield more information.

How much information do 258 features really contain?

It seems that introducing some geek-speak is unavoidable at this point. My apologies to the less mathematically interested readers. I will return — after a brief plunge into the weeds — to an explanation that is more plain-English.

Principal Component Analysis (PCA) tells us how information* is distributed between the features in a data set. PCA "reshuffles" the features using a mathematical method called Singular Value Decomposition (SVD). The features in our data set are often interrelated to each other in complicated ways. One factor might depend on several others. They contain lots of redundancy!

SVD unwinds all of the features into an "abstract" feature set. The members of this "abstract" set are unrelated to each other.  SVD also tells us how much information each "abstract" feature bears.

This is the point: Our data sets all contain a mixture of true, value-added information and noise. The actual information is usually spread across lots of different features. So is the noise. Data scientists — when they build analytics models — struggle to separate the real information from the noise. When the real information is spread across too many features, it's easier for it to become buried in the noise. The more features the information is spread across, the more diluted it is.

Figure 1 illustrates.  (I used the R open source statistics tool to produce Figure 1.) It shows how the information* is distributed across the 70 "abstract" features. Two of those features contain 95% of the information about how the 923 occupations are related to their associated skills. The remaining 68 features — containing only 5% of the information — are mostly noise.

Figure 1 — Two of 70 "abstract" features from the O*Net occupational skills table contain 95% of the information about how skills are related to occupations.

What does this mean? Most of the 70 "original" features in the skills data set are highly redundant. BLS provides a summary of the Holland model's history. They recently issued a newer version than the one I studied here. This release contains new 126 occupations.  

Are they adding new features to the occupations? I have not done that analysis. New features may be attractive from a career consulting point of view. Figure 1 however suggests that addition of new features should be guided by caution. New features should primarily add new information that is not duplicative of the 258 that are already in the data set.

But, Figure 1 just shows us the Skills features that BLS uses to characterize occupations. What about the rest?  What happens when we put them all together? Figure 2 gives us the answer.

We saw in Figure 1 that two abstract features contained 95% of the skills information. We add Abilities, Knowledge, and Interest in Figure 2. We might expect that four groups of factors would lead to eight information-bearing abstract features. This is not the case! 

Figure 2 — When we combine Skills, Abilities, Knowledge, and Interests, 95% of the combined information is contained in just four of 258 "abstract" features. 

The marginal value of information added by increasing our occupations characteristic data set from 70 factors in Figure 1 to 258 factors in Figure 2 is just two additional information-bearing abstract features. The Abilities, Knowledge, and Interests feature sets don't tell us much more than we know from Skills alone. 

We increase our feature set by a factor of about 3.7. We only double the number of information-bearing features. This suggests we have reached some form of diminishing returns. 

Leave the kitchen sink out of your data strategy for analytics.

In last week's installment I introduced some rudimentary concepts from Information Economics. The value businesses get from information is related to the opportunity that information gives to take more profitable actions. 

By extension, does data bring value? I illustrate above that adding data only increases value if the new data tells us something we don't already know. From an information point of view, Abilities, Knowledge, and Interests don't tell us much more about the 924 occupations in BLS' O*Net database than the Skills features alone. Adding those three data sets increased our feature set from 70 to 258, but only gave us two additional information bearing features.

But handling data is accompanied by cost! Technology vendors call our attention to the plummeting price of storage. Storing information is becoming increasingly cheap. Extracting value from data?  Not necessarily so.

The big data buzz might tempt us to expect that cost-free opportunities exist to extract value from data. We routinely are regaled with stories about Internet companies' use of data mining to illuminate many of the most intimate details of our lives. Google, Facebook, Yahoo!, and others accomplish this through strategic collection of data elements that introduce new information. They appear to have borrowed knowledge from the forensic computer security business to disinter our secrets.

What's the bottom line? Organizations seeking to extract value from data need to focus on value in their use of data. Begin with an Information Economics philosophy. Ask questions like:

• What information will give me improved economic value?
• How much value-bearing information is contained in the data that I already have? and
• How much will handling the data cost to the new value-bearing information that I need?

Next installment:  Elements of the cost to extract information from data.

* I use the term "information" here imprecisely. "Singular values" produced by SVD are not information, technically speaking. They are related to information content. I intentionally commit this error in terminology for convenience of discussion.

© The Quant's Prism, 2014

Information Economics: The Foundation of Business Analytics.

The technology industry often produces multiple, not-altogether-consistent definitions of the latest "hot" thing. Business analytics is no exception.  Confusion can be the result.

Hence, the intensity of my focus on definitions. Having previously defined "data science", I now drag readers through an exercise in defining business analytics.  At the risk of appearing obsessively compulsive, I repeatedly emphasize the business context.

The charter statement for this blog emphasizes the study of the application data science to business problems.  I seek to apply the scientific method to its practice.
"Data and statistical methods" have become inseparably associated with the "how" of business analytics.  I want want to dig deeper.  

Science is about answering "why." Since business is the domain of interest for business analytics, we should look to economics as a candidate foundational science. In the following, I:

• Make the case for foundation for business analytics in economics;
• Introduce a specialized domain of economics on which business analytics is based; and
• Provide a simplified illustration of its use.

Why, reader, should you care about these things? Big data and business analytics are subjects of many bold claims regarding their transformational abilities.  Some of these claims are valid, and some are not.  I seek here to help you separate the science from the alchemy.

What is business analytics?

Service oriented architecture (SOA) — a source of significant tech-industry buzz during the last decade — provides a case study in definitions.  Distinct definitions appeared to arise for each stakeholder class. The Open Group — a non-profit organization promoting open standards for technology view — offers two definitions of SOA.  Software vendors tend to emphasize the key technology components.  For SOA, those are an Enterprise Service Bus (ESB) and a services registry.

Merrifield, et al,¹ identified the business payoff for a SOA approach to strategic technology management. SOA promises a cost-effective approach to mass customization of information technology. They describe a SOA planning method. But they did not explicitly define the practice.

Enterprise architecture (EA) provides another example.  I like Gartner's definition for two reasons:

• The veracity of its source (i.e., Gartner said it); and
• Its focused on "enterprise" in the business sense, independent of the technology.

The technology community captured the EA term to connote the architecture of the IT infrastructure — for either an organizational enterprise or for an individual system. Technology consulting group Forrester Research invented a new term, Business Architecture, apparently in response. Ross and Weill² established the need for a business-centric definition.

Why this circuitous path?  I want to make the case for an economics foundation for business analytics. Getting definitions right is important to my case. Statistical (and deterministic) models, modeling tools, and enterprise information management technologies are "how" business analytics is done.  Science seeks to answer the question, "Why"?

The important point here is that we focus business analytics on answering questions that lead to measurable, net-positive business outcomes.  We seek a scientific underpinning from which to achieve this objective.

What, then, does this mean for the discipline of business analytics? I continue with the pattern of a business-centric perspective. I also want to define it as precisely as possible. Wikipedia, the reflexive "go to" source, offers a pretty good definition:

Business analytics (BA) refers to the skills, technologies, practices for continuous iterative exploration and investigation of past business performance to gain insight and drive business planning. Business analytics focuses on developing new insights and understanding of business performance based on data and statistical methods.

This definition borrows from Bartlett³ (whom I have yet to read, but have added to my Kindle wish list).  Davenport⁴ defines the payoff — "improving performance in key business domains" — without explicitly providing a definition.

So, business analytics is about quantitative characterization of business performance. What then is business about?  Those who have sat through an MBA program might observe that the majority of the curriculum is derived from economics and its applications. An economic foundation for business analytics therefore seems reasonable.

Hence, an economics grounding for business analytics.  I preserve here the distinction between business analytics and econometrics.  The two disciplines use many of the same tools. Business analytics focuses however on a distinct organization.  It arguably constitutes a subset of econometrics.

The foundation of business analytics

Information economics is business analytics in its must fundamental form. Information economics is the science of assigning economic value to information.  It combines principles from the following disciplines:

• Game theory, by which economic transactions are defined and modeled;
• Information theory, with bases both in engineering and psychology disciplines; and
• Microeconomics.

Figure 1 illustrates. 

Figure 1 — Information Economics resides at the intersection
four more familiar disciplines.

Practitioners of Info Economics employ a clearly defined toolset.

• Information modeling,⁵ borrowed from information theory, precisely represents the distribution of elements of information among participants in an interaction;
• Game theory contributes pattens for archetypical transactions between participants in an information exchange; and
• Microeconomics provides bases for economic valuation of elements of information involved in a transaction between counterparties.

The presence of uncertainty in economic transactions introduces probability theory as well.

Information economics provides the foundation for many well known theories about the operation of financial markets.⁶  The interplay between bid-ask prices in a financial exchange, for example, telegraphs considerable information about counterparties' intentions and abilities without explicitly "showing their hands."  The Efficient Market Hypothesis finds partial justification in Info Economics.

Marketing economics is replete with examples.  Applications occur of course in other business disciplines.  For example, information economics can inform investment decision making.  Its principles can also guide what aspects of operational cost and efficiency are most worthy of measuring.

As an aside, the similarities between information economics and real options⁷ are striking. Real options theory assigns economic value to flexibility in making investment decisions.  I leave that discussion for a future installment.

A simple, contrived illustration of assigning economic value to information†

My illustration here is based on the "Elementary Game,"⁸ one of the simplest models from game theory. It resembles the Binary Symmetric Chanel (BSC). I first saw the BSC in communications courses while studying electrical engineering.  Info economics and communications theory (the engineering variety) share roots from information theory. That their toolsets resemble each other does not surprise me.

Figure 2 — a BSC illustration — gives us a passable representation of the "Elementary Game." (Note:  Texts in communications theory⁹ and info economics share the "Alice" and "Bob" notation.) The a priori events appear on the left-hand side.  There is some probability of either of two events occurring.  "Alice" initiates one event or the other.  

"Bob" receives a "signal" indicating — with a probability p that it is correct — which event "Alice" effected.  He therefore views the event from an a posteriori perspective.  Based on knowledge of the a priori probability of what "Alice" did, the probability that the signal is correct, and the cost/benefit of either of two resulting courses of actions, "Bob" must decide what his optimum next step is.

Figure 2 — The "Elementary Game" from game theory resembles the Binary Symmetric Chanel (BSC), a basic building block of communication theory.  (Source: Wikipedia, http://en.wikipedia.org/wiki/Binary_symmetric_channel)

Let's see the "Elementary Game" in action. Say that I'm a BMW dealer. I operate in a market that generates 10,000 sales annually.  I capture an average of 1,000 of those sales — or a 10% market share. Sales produce an average of $50,000 in revenue.  I have traditionally used mass media — broadcast and newspapers — for my advertising.

Let's now say that I can identify decision factors — information — that influence buyers' decisions about whether and from whom to make a purchase of a new car in my market segment.  These factors might include:

• Capacity to make the purchase (e.g., disposable income);
• Brand preferences; and
• Age of current vehicle;

among other factors. I can use this information as the basis for a targeted advertising campaign that — with probability of 25% — increases my market share to 15%. Assume that (in order to keep this example simple) the targeted ad campaign costs the same as the mass media campaign.

How much is this information worth?  Information economics defines the value of information as:  "...the increase in utility from receiving the information and from optimally reacting to it."⁸  So, without the information I can take a course of action leading to one outcome — a specific revenue level in our case.  Given the information, I can make a decision to pursue an alternative course of action.  This alternative leads to a different outcome. I characterize my two alternatives using the same measure.

The increase in utility in our example is change in revenue realized from a targeted ad campaign. From elementary probability theory,

ΔRevenue = Pr{ΔSales} × ΔSales × Average revenue/sale
             = 25% × 500 × $50,000 
≈ $6,250,000.

Information that can — with probability 25% — increase my market share by from 10% to 15% is worth about $6 million to me!  This trivially simple illustration demonstrates the power of Google's business model.

Information economics at work

I illustrated a scientific approach — based on Information Economics — to assigning value to two specific elements of information in a specific business context. These essential elements of information are:

• What change in market share might I be able to effect with a targeted ad campaign; and
• What is the probability that my targeted ad campaign will produce that result.

This gives me an economic value of those two information elements.  I apply business analytics — data and statistical methods, tools, data scientists — to obtain the answers to these specific questions.  If the cost of getting this information is less than its economic value, then applying business analytics here yields a net-positive economic benefit.

This illustrative example admittedly oversimplifies things. Business decision makers should base their decisions on a range of probabilities. Few business questions lead to discrete, binary answers.

So, what does all this mean?  First, information economics provides a scientific approach to business case analyses for business analytics initiatives.  The value of business analytics is measured by its economic returns.  We now have a rigorous approach to determining the "goodness" of business analytics initiatives.

Second, this leads to criteria for a strategy for adoption of business analytics by organizations.  Lavalle, et al, advise would-be data-driven organizations to, "Start with questions, not data!"¹⁰  Successful adopters of business analytics as a foundation for decision making keep a laser-like focus on:

• Business outcomes; and 
• The questions that lead to them. 

This also implies a gradual, evolutionary approach to adoption.  But more on that, later.

Next installment:  Is more data always better?

Note:  Missed my cadence last week.  A short-notice proposal turned a slack week into a frenetic one.  But back in the saddle again, this week.

¹ R. Merrifield, J. Calhoun, and D. Stevens, "The next revolution in productivity," Harvard Business Review, June 2008,  http://goo.gl/Y58xqm.

²J. W. Ross and P. Weill, Enterprise architecture as strategy, Boston:  HBR Press, 2006, http://goo.gl/B7J5P8.

³ R. Bartlett, A practitioner's guide to business analytics," McGraw-Hill, 2013, http://goo.gl/o6dTOS.

⁴ T. H. Davenport, J. G. Harris, and R. Morison, Analytics at work, Boston:  HBR Press, 2010, Location 112, Kindle Edition, http://goo.gl/olZkKm.
⁵ L. Samuelson, "Modeling of knowledge in economic analysis," Journal of Economic Literature, June 2004, pp. 367-402.
⁶ M. K. Brunnermeier, Asset pricing under asymmetric information, London:  Oxford, 2001, http://goo.gl/7IMFDv.
⁷ See, e.g., M. Amram and N. Kulatilaka, Real options, Boston: HBR Press, 1999, http://goo.gl/6Bswjk.

⁸ M. Bütler, Information Economics, New York:  Routledge, 2007, p. 42, Kindle Edition, http://goo.gl/1zZKQ1.
⁹ see, e.g., B. Schneier, Applied cryptography, 2nd ed, New York:  Wiley, 2001.

¹⁰ S. Lavalle, et al, "Big data, analytics, and the path from insights to value," MITSloan Management Review, Winter 2011, pp. 21 - 31, http://goo.gl/8RSn5H.

† This example is purely fictional.  Any resemblance to experiences by actual BMW dealerships is purely coincidental.

© The Quant's Prism, 2014

The "Science" of Data Science.

The inaugural installment of this journal described a broad vision related to the study of "data science." I attempt here to focus that ambition somewhat. Attempting to define the term "data science" is the objective of the this discussion.  

Most importantly, I want to assert a point of view that emphasizes science. I also approach data science from the point of view of a business professional. I am interested in applications of data science to deepening insight into strategies and operations of businesses and other organizations.

As an aside, I found this task more difficult than expected. A recent NY Times op-ed piece asserts that "Writing Is a Risky, Humiliating Endeavor." My experience here may corroborate that author's point of view.

The challenge with defining "data science" is that it is intertwined with a number of related IT-industry terms including, but not limited to:

• Business intelligence;
• Analytics and business analytics; and
• Big data.

A number of Internet tools attempt to measure the intensity of attention given to topics.  Google Trends reports relative rates of search-engine queries.  The graphic below reports the relative frequencies for our four terms of interest.  We see that the term analytics attracts the lion's share of the interest.

"Data science" is practically lost in the noise. The growth rate for "data-science" queries appears relatively steady. Seeing total-volume statistics would be interesting.


A number of definitions of "data science" have been attempted:

• Strata's Mike Loukides published What is Data Science? through O'Reilly.  This meandering narrative covers technologies and anecdotal instances of their use. A definition of data science is not concisely presented.
• Gartner performed a text-analytics study identifying the core competencies of a data scientist as data management, analytics modeling, and business analysis. Gartner extends this list to "soft" consultant-related skills.  We find here an indirect definition of data science — in terms of its primary practitioner.  
• Wikipedia provides a definition represented as an intersection of computer science, applied mathematics, and subject-domain expertise for a specific field.
• Forbes contributor Gil Press, in "A very short history of data science," observes 52 years of history of the use of the term.

Regrettably, no definition suitable to our purposes conspicuously presents itself.  I therefore presumptuously undertake to proffer my own. 

What is Data Science?

We seek a business-focused definition of data science.  We also emphasize the scientific aspect. Business analytics captures one aspect of data science.  Data science can moreover be applied to "big data."

My approach is inspired by a 2008 Gartner report, "Gartner Clarifies the Definition of the Term Enterprise Architecture."  I address each of the points in the outline used for Gartner's definition.

What it is: Data science is the application of the scientific method to extract actionable insights from diverse sets of business information. 

What the scope is:  Data science prescribes systematic, reproducible methods to the entire information lifecycle from information source to information consumer.  This includes mathematics, data visualization, information management, and business analysis.   The breadth of data science's span transcends business domains of strategy, operations, finance, and logistics. 

What the result is:  The application of data science to business information leads to the best achievable information on which to base a specific business decision or action. Its results are organized and presented in a manner specifically designed for the information consumer.  They also contain indications of the degree of confidence to which the consumer should assign to them.

What the benefit is:  Consumers of information produced by data science receive the best achievable, actionable information specific to high-priority business questions. This results are provided the minimum expenditure of resources compared to less-scientific approaches. The scientific method leads to application of the most-appropriate mathematical and information-management methods to answer specific business questions given the available data.

How is Data Science related to the Scientific Method?

My preference for the term "data science" over "analytics" or "big data" is grounded in the prominence of science.  "Applying classic scientific methods to the practice of management¹" is one of the key promises offered by the movement encompassing data science, business analytics, and big data. "...The ultimate goal of data science is improving decision making, as this generally is of paramount interest to business.²" Improved business decision-making leads to:

• Improved predictability of decision activities;
• Reproducibility and transparency in the decision-making process; and
• Precise separation of uncertainty into aspects that can be mitigated and those that cannot.

Data science — including big data, business analytics, and business intelligence — can never completely remove uncertainty from making decisions. It does separate resolvable uncertainties from those that remain — to varying degrees — "known unknowns."  

So how do business-focused data scientists apply the scientific method to business analytics?  Business analytics thought leaders describe high-level approaches in leading business journals.  

• LaValle, et al,³ describe an incremental, "outside-in" approach to an analytics-centric enterprise strategy for IT.
• Davenport⁴ lays out a six-step "procedure" by which managers keep their data scientists "honest".

This list is by no means exhaustive.  They are consistent with analysis methods in which data scientists are indoctrinated during their educations.

I turn to Pirsig⁵ — an admittedly quirky source — for an accessible summary here.  Pirsig summarizes the scientific method in four steps:

1. State the problem in terms that are no more than you are positive that you know;
2. Formulate hypothesis of candidate causes for the problem;
3. Design experiments to test the each hypothesis in isolation;
4. Interpret the experiment results in terms of whether the hypothesis is proven or refuted; and
5. Update the candidate hypotheses and return to step 3.

This is similar to Davenport's⁴ six-step "procedure."

This rigorous, systematic approach is necessary and sufficient to achieve the fundamental objective of data science in business decision-making:  Isolating uncertainty factors into those that can be resolved and those that remain uncertain.  Cutting corners leaves residual uncertainty.

Practicing analysts who may have read this far may ask, "But, what about unsupervised learning?"  I will address unsupervised learning in depth in a later installment dedicated to the topic.  Suffice it to say, for now, that unsupervised learning provides a source of candidate hypotheses.  Each resultant candidate hypothesis must be scientifically tested as described above.

Next Installment:  The economics of information.


¹ "Big data: The next frontier for innovation, competition, and productivity," McKinsey Global Institute, McKinsey & Company, 2011, p. 98,  http://www.mckinsey.com/~/media/McKinsey/dotcom/Insights%20and%20pubs/MGI/Research/Technology%20and%20Innovation/Big%20Data/MGI_big_data_full_report.ashx.
² F. Provost, T. Fawcett, "Data science and its relationship to big data and data-driven decision making," Big Data, Mary Ann Liebert, Inc., February 13, 2013, p. BD 53, http://online.liebertpub.com/doi/pdfplus/10.1089/big.2013.1508.
³ LaValle, Steve, et al, Big Data, Analytics and the Path From Insights to Value, MITSloan Management Review, Winter 2011, http://sloanreview.mit.edu/article/big-data-analytics-and-the-path-from-insights-to-value/
⁴ T. Davenport, "Keeping up with your quants," Harvard Business Review, July-August 2013, http://hbr.org/2013/07/keep-up-with-your-quants/ar/1
⁵ Pirsig, R. M., Zen and the Art of Motorcycle Maintenance, Harper-Collins, 1974, (Kindle edition 2009), http://goo.gl/si1ayP

© The Quant's Prism, 2014

Introduction: Motivation and Charter

Charter

I dedicate this journal to promulgating and testing ideas about mathematics and its role in society and the economy. The popular media seems to pay Mathematics much attention these days.  Imaginations about the potential of the closely related field of artificial intelligence have also been reignited.  IBM's Watson project makes Stanley Kubrick's 2001: A Space Odyssey HAL 9000 character seem almost within reach.

I use this venue to contribute my verse — to borrow from Whitman — to the data science cacophony. Installments will address topics including:

• Consideration potential and limitations of data science to decisively contribute to business and public policy;
• Summary and critical analysis of topics of current interest in the media; 
• Discussion of approaches to applying data science to practical problems;
• Review of noteworthy literature germane to data science; and
• Illustrative results from my own computational tinkering.

Providing a dispassionate view of selected topics related to mathematics will be a successful outcome of this project.  

I intend to attempt weekly installments. One installment a month will be a quantitative case study. Openly available data sources provide rich troves of opportunities for curious analysts.  The first installments will use labor-economics data from the bureau of labor statistics.

The remaining weekly installments will cover "softer" topics. The list of "soft" topics contains fifteen entries. The order is still under consideration.

Motivation

Why write about Mathematics and its role in society and the economy?  A couple of reasons come to front of mind:

1. I love mathematics.  Insight into a nuance satisfies like few other things. Math represents to me The Spiritual; The Mystical; A Glimpse of the Divine.
2. Math has suddenly become cool!  The popular media recently produced shows like Numb3rs.  Harvard Business Review called data science "The sexiest job of the 21st century."   Yes, we still have Big Bang Theory.  But it's a long way from Steve Urkel.

So how can somebody feels this way about Mathematics provide a dispassionate view?  I can because I am the recipient of a scientific education. The scientific calling demands passionate dispassion.

My experience and education also equip me for dispassionate exploration.  For example:

• I follow the pattern of a one-time consultant and mentor.  R.C. Hansen published a widely circulated paper about the fundamental limitations of his field, antennas.  The grip of this arcane, narrow discipline upon him was sufficiently powerful to drive him to edit or write the definitive reference in his field not once, but twice.  The first appeared in 1985 and the second in 1998.  (A copy of each collects dust on my bookshelf.)  This commitment sharpened his objectivity.
• I view the limitations of mathematics as an attribute of its beauty. Applying mathematics to business and public policy is fundamentally grounded in information theory. Cover and Thomas open their Information Theory text with presentations of the Information Inequality and the Data processing Inequality.  These fundamentals constrain data science to the same extent that gravity and drag impose bounds on aerodynamics.

But, the popular media appears to assign limitless potential to mathematical computation perpetually increasing due to the inevitable advance of Moore's Law.  The Gartner Group annually publishes a variety of "Hype Cycles."  Many within the IT industry will be familiar with this methodology.  A hype cycle represents a technology's lifecycle.  Similar methods appear elsewhere in the business literature (e.g., see Moore, HBR, July 2004).

Figure 1 shows Gartner's 2013 installment of its "omnibus" Hype Cycle for Emerging Technologies. (The entire volume is available for purchase at a price accessible to many medium-sized and large businesses.)  "Big Data" appears prominently at the "Peak of Inflated Expectations.  Big Data is frequently applied to the application of data science.

Figure 1 — The Gartner Group's 2013 installation of "Hype Cycle for Emerging Technologies." Source:  http://goo.gl/a4xlEY.

What does this mean for Data Science?  It is presently the boom phase of the boom-bust cycle followed by many industries. Gartner forecasts that "Big Data" will reach the "Plateau of Productivity" stage of industry maturity in five to ten years.  Between now and then, many dreams will be smashed, hearts broken, and fortunes lost.  The illusion that any analytical problem can be solved given enough CPU clock cycles will be shattered.

How does one get from now to then?  Two obvious paths present themselves:

• We can hop on the roller coaster, raise our arms above our heads, and scream at the top of our lungs; or
• We can practice passionate dispassion.

My preference ought to be obvious by now.