How steep is your learning curve? On Analytics and Mentors ...

Having a mentor is the number one factor in increasing the steepness of your personal learning curve. So says my oldest, Garik, a Park Scholar at North Carolina State University (class of 2012), during a discussion he recently had with the incoming Park Scholar class of 2019.

learning-curveTo accept the value of mentoring first requires one to understand the centrality and importance of the learning curve. Garik asked the students to imagine plotting the characteristics of two people on a simple X-Y axis.  Person A comes to the game with only a moderate amount of resources at their disposal, but importantly, also a relatively steep learning curve, such that a plot of their capabilities has them crossing the Y-axis at an intercept of 1 and with a slope of one-half.  Person B, in contrast, has much greater resources at their current disposal:  time, talent, smarts, money, education, experience, etc …, but for whatever reason, has a shallower learning curve, such that their plot on the graph intercepts higher up the Y-axis at 2 but with a shallower slope of only one-quarter.

Unless you think you’re going to die before the two lines cross, you’d of course be better off as Person A. Based on his domestic and international experiences as an undergrad and grad student, as a researcher and an employee, and as part of two start-ups (so far), Garik’s conclusion is that, while there are several factors impacting the steepness of that learning curve, none is more important than that of having chosen good mentors.

Businesses can be said to have learning curves as well, and my discussion with my son got me to thinking about what factors would have the greatest bearing on organizational learning curve steepness. Read More »

Post a Comment

Your information supply chain

Viewing data as an asset implies there are benefits to taking a supply chain approach to data management. It’s not just inventory that needs to be at the right place at the right time in the right format and quantity. An end-to-end information supply chain approach, from sourcing/acquisition through transformation and storage, to end users, analytics and insights, allows you to keep your focus on the business problems to be solved, and avoid having that ‘big honking data cube’ become a bottleneck instead of an enabler.

Here are a few best practice supply chain mindsets to consider applying to data management:

  • cadworx2d3dInventory turns / Data turns: How fast can you acquire and get the needed data into the hands of the decision makers? The whole point of analytics, BI and corporate performance management is to make better decisions faster, and how you structure your information supply chain will have a big impact on that “faster” part. This could mean anything from self-service BI to data visualization to better and faster data prep and data quality procedures. Critical decision support data delivered too long after the problem surfaces is like not having the inventory you need in the stores until the holiday shopping season is half over.
  • Analytics / Decisions at the edge: Taking “faster” to the extreme can sometimes mean taking action on the data BEFORE you store it, acting on streaming data via event stream processing, which has applications in financial services (e.g. credit scoring), cybersecurity and fraud (e.g. detecting unusual network connection patterns), quality (e.g. process and product quality control), and asset maintenance (e.g. sensor data from critical equipment). Whether it’s humans or machines making those decisions at the edge, real-time and near-real-time decision support capabilities are becoming competitive differentiators across a number of industries.
  • Visibility / Control Towers: Reacting quickly to fluctuating demand requires visibility into your entire physical supply chain, from what’s in which store or warehouse, which production lines are down for maintenance, and which suppliers have additional capacity at the ready. In the same way, better business decisions means having immediate access to ALL the relevant data – a worthy integration challenge for both inventory and data, with a commensurately worthy business outcome.
  • ABC inventory / data classification: Not all inventory is equally important, a maxim understood by every supply chain professional. “A” inventory is high in value but not necessarily volume, warranting tighter controls and monitoring than low-value, high volume “C” material. Your data can likely be similarly categorized, with revenue, cost, employee and customer data no doubt in the “A” category, with perhaps production, quality and transaction details falling lower in priority. Such a classification approach, based on what data is used most often in the most critical decision support processes can help your prioritize your data quality and myriad other IT activities and investments.
  • Data Management for Analytics: Designing a physical warehouse that maximizes the efficiency of receiving and storage is not necessarily the best overall approach to supply chain management, where access to raw materials and WIP at the right place and time on the factory floor is more critical to meeting cost, revenue, customer satisfaction and business goals than simply minimizing storage and handling costs. Likewise, a data warehouse built to minimize data storage cost may make it difficult for business and decision support users to access what they need, in the format they need for rapid analysis and insight. This gets back to point number one above about data turns: it’s not about how fast and cheaply you can get data into the warehouse, it’s about how quickly you can turn that data into valuable insights across the entire information supply chain.

Read More »

Post a Comment

Painting with big data analytics

Seurat-La_Parade_detailBig data, by which most people mean Big Volume, doesn’t get you very far just by itself, but with the addition of Big Variety and analytics, now you’re talking. In fact, most organizations who are making headway into capitalizing on their data assets now refer to the process as "big data analytics" – a combination of data storage and management, data integration, and analytic tools and techniques.

I have previously made the case that the value in big data stems primarily from its Big Variety, and now I want to put that variety into its proper context of related data volumes and analytics for insights.

The potential for getting this combination of Volume + Variety + Analytics right can perhaps be best illustrated by the findings of a U.S. Department of Defense review committee on the September 11, 2001 attacks. What the committee found was that essentially all 19 of the hijackers could have been linked to each other and to the pending attacks via just seven properly targeted mouse clicks through existing public/government databases: Read More »

Post a Comment

Marketing analytics lessons from the KGB

“Half the money I spend on advertising is wasted, the trouble is I don't know which half.” ~ John Wanamaker, U.S. department store magnate and merchandising / advertising pioneer.

I’m not going to claim that I can pinpoint exactly which half of your marketing dollars are wasted in the space of this post, but I am going to illustrate that basic analytic techniques are available that can considerably narrow down the range of uncertainty and provide actionable insights for your marketing efforts.

KGBOur story begins with a fascinating article that surfaced last week by Jonathan Haslam, professor of the history of international relations at Cambridge University, the subject of which was how, during the Cold War, was the KGB able to so easily and readily identify undercover CIA agents?

The Soviet efforts were so successful that the head of the KGB counterintelligence group, Yuri Totrov, was known within CIA circles as the “Shadow Director of Personnel” on account of how much he seemed to know about the foreign posting of CIA agents. How he was able to unmask and compromise entire intelligence networks was the subject of much handwringing, debate and speculation, the leading candidate being a highly placed mole within the Agency. What other explanation could there possibly be, right?

Wrong. Read More »

Post a Comment

What is it like to be a customer?

bat2To paraphrase Thomas Nagel’s famous 1974 paper on consciousness, “What is it like to be a bat?”, I want to instead ask the question, “What is it like to be a customer?” Nagel’s argument was geared at refuting reductionism - the philosophical position that a complex system is nothing more than the sum of its parts. Such a materialist approach omits the essential components of consciousness ("emergent properties" we would say today): an actor with motives and feelings and a personality. We typically approach the customer in the same fashion – our hypothetical target consumer is typically nothing more than the sum of our data and demographics combined with our own products and services.

I want to digress for a moment to illustrate and highlight this important point about “being like something”. What is it like to be you? Read More »

Post a Comment

Big Model: The necessary complement to big data

With all the hype over big data we often overlook the importance of modeling as its necessary counterpart. There are two independent limiting factors when it comes to decision support: the quality of the data, and the quality of the model. Most of the big data hype assumes that the data is always the limiting factor, and while that may be for a majority of projects, I’d venture that bad or inadequate models share more of the blame than we care to admit.

stone-balanceIt’s a balancing act, between the quantity and quality of our data, and the quality and fit-for-purposeness of our models, a relationship that can frequently get significantly out of balance. Or more likely, complete mismatches between data and modeling can crop up all over our organization. In one instance we may have remarkable models starved for good data, and on the other hand, volumes of sensor or customer data sit idle with no established approach to exploration, analysis and action.

This imperative to balance the data with the model reminds me of an espionage story from WWII. Read More »

Post a Comment

Visualization – Worth a thousand words

Why visualization? Several reasons, actually, the most compelling being that sometimes visualization literally solves the problem for you.

I remember an exercise in eighth grade English class where we were asked to describe, in words only, an object set in front of us with sufficient clarity such that our classmates, sequestered outside the room, could accurately draw the object from our written description. The object was a bowtie shaped set-top UHF antenna.

bike2The exercise was a disaster. Which of course was the objective, at least from the teacher's perspective, who was attempting to demonstrate how difficult clear, comprehensible writing can be. From our perspective, however, all we could focus on was what idiots the recipients of our written descriptions must have been. “Why did you draw the loops at right angles to each other when I clearly indicated they were in the same plane as the base?” Looking back, had I been clever enough, I realize now that I should have “cheated” and used a typographical approach to illustrate the object diagrammatically with my otherwise 'descriptive' words.

Live and learn. But what did I learn? Read More »

Post a Comment

Why build models?

why2We are all modelers.  Whenever you plan, you are building a model.  Whenever you imagine, you are building a model. When you create, write, paint or speak, you first build in your head a model of what you want to accomplish, and then fill in the details with words, movements or other actions in order to realize that model.

Models work via a three-part structure:  Input, Mechanism, and Output.  When we use models, we are generally confident in only two of the three stages, and we use the process to determine the unknown stage.

The most familiar construction is where we know, or have confidence in, our inputs and our mechanism (the mechanism being the rules or algorithm that generates output from inputs), which we call “prediction”.  We use this structure to predict, or forecast, a wide variety of outputs, from tomorrow’s weather to next month’s sales to next year’s election.

But prediction is not the only available model structure.  Read More »

Post a Comment

Big data demands Big quality

data qualityUnlike data acquisition, which can accumulate exponentially, we generally address data error correction on an exception basis, using manual, linearly-scaled resources.  We cannot possibly scale manual data correction to keep up with our increased data volumes, which means we must automate our data quality processes with tools at least as robust as our data collection and storage resources.  We cannot afford to scale up the human resources that today correct perhaps 100 customer name, address, part number or shipment date errors per week, to handle thousands or tens of thousands of such errors.  Our only alternative is to automate and catch / fix those errors up front.

In a previous post (“Big Silos”) I posited three primary sources of big data:  visually dense (e.g. video, satellite), temporally dense (e.g. audio, sensor), and transactions (e.g. POS, SKUs).  I would like to amend that classification to now include a fourth primary source of big data that I had initially overlooked: unstructured data, including social media data.

80% of corporate data is unstructured, making it perhaps the single largest potential source of big data.  As we start to process, structure and store that unstructured data, through techniques such as text analytics and content categorization, we need to remember to apply the same data quality standards as we do to more traditional transactional data. Read More »

Post a Comment

Homo habilis to Homo sapiens – From tool user to data user

Are robots going to put us all out of work?  That seems to be the question on every click-bait economic article’s mind.   On this particular issue I’m taking a firm position:  NO! Emphatically, NO!  Robots are not going to take our jobs in any permanent sense that would entail a total reworking of basic human economics.

How can I be so certain?  Because we’ve been through this before on an even greater scale and not only survived but largely thrived.  I am of course talking about the industrial revolutions, both first and second, and whatever atomic / space / computer / information revolution we might be in right now.

tools1-olduvaiWhen speaking of the first and second Industrial Revolutions, we don’t refer to them as “robots” but instead call them machines and equipment.  Still, they are all kith and kin, part of the larger category we refer to as “tools”.  And when it comes to tools, we’ve been at this for 2.8 million years, so chill out everybody.  In fact, the first in line in the genus ‘homo’ was Homo habilis, “handy man/woman”, or ‘tool user’.

However, it’s not the Olduwan flaked stone tools that I’m entering into evidence here, but instead the machinery of every shape and kind that filled the first factories and furnaces of the industrializing West, displacing and disrupting the craft guilds and the farmers, but not resulting in the economic extinction of the species. Prior to the start of the Industrial revolution roughly 80% of Europe’s economy was agrarian.  By 1900 agriculture still accounted for 40-50% of the West’s GDP, and by WWII it was down to 15%.  Today those numbers hover around 2% for Europe and less than 1% for the U.S.

When it comes to the trade guilds, yes, the coopers, cobblers and colliers are long gone, and slightly more recently we’ve seen the demise of lamplighters, ice cutters, milkmen, bowling pin setters and switchboard operators.  And yet here we are. Read More »

Post a Comment