Tag Archives: analytics

Analytics – Trust the Data and Your Gut?

So earlier this year I entered an analytics contest and I won!

WinnerWinner

TDWI sponsored the contest, using analytics, to predict the opening weekend for Star Wars: The Force Awakens. HUGE, was not an acceptable answer, so I started to formulate how one would predict such a widely variable number. Here was my thought process, based on the data available and in the end, what I felt in my gut.

I first looked at the past openings for the Star Wars franchise; removing the special editions and re-releases, Using the number of theaters each movie opened in, and what the average ticket price was by year. I also approximated the number of tickets sold ( for an average theater size).

Looking at comparable films (i.e. “Blockbusters”) for this year, I estimated the number of theaters (about 4200) that SW:TFA would open in, and this year’s average ticket price (> $8.15 a seat).

This brought me to a figure of about $176M, which I felt really low. A small sample size of only 6 movies will do that…

Next, I looked at opening weekends for this year, with Jurassic Park being the largest at just over 200 million. No way Star Wars isn’t beating Jurassic park. No way. So where would the difference come in and make this opening remarkable, considering the Disney Machine of Marketing plastering ST:TFA on EVERYTHING.

The last tweaks in the numbers were guesses really.

  • Cost different between standard and 3D; I guessed more 3D views would push the average ticket price up.
  • Number of repeat viewers, guessing 40%, maybe higher depending on how good it was… but I went with 40%
  • Social Influence, if it was good people telling their friend may push the box office $ up. This would be negligible at best
  • Generational “Drag-Alongs” (Gen X and Gen Y, who are dragging their kids along for nostalgia). This would only be first views, not multiple in most cases.
These last 4 were “thumb in the air” guesses to be totally honest. Analytics got me partly there, my gut got me really close!
My final prediction/guess was $247,575,500. Less than 1% from the actual weekend opening draw.
Considering the small sample size I had to work with of data, and the selective nature I used to create my data set, I got lucky. But the predictive/what-if piece of the analysis is really what got me close. My thumb in the air variables were pure gut instinct, but being able to model the potential future was an important factor in getting me to my final entry.
Boiled down, it was basic BI:
  • What did the franchise do in the past?
  • What have competitors and others in the industry done recently?
  • What marketing/awareness efforts are in place and how effective are they?
  • What does the customer’s sphere of influence look like?
  • What does brand strength do to the bottom line?
Is there a lesson here? Probably not, but it reaffirms to me that sometime when we try to over complicate things we can lose sight of what is real. Keep it simple, trust the numbers and let people apply their gut.

 

Building a Better Analyst

I participated recently as a panelist in a Business Intelligence Summit at the University of Akron as they had identified a gap that their graduates had when entering the workforce; they had little or no analytical skills when it came to data. The faculty was exploring the idea of adding business and data analytics classes to the curriculum for business majors to fill that gap. We were there to discuss Business Intelligence and Analytics and impress upon them the opportunities that exist in the job markets, the fields those jobs were in and the skills needed to compete. Christina Rouse, Chief Architect for Incisive Analytics, presented a series of slides that explored the skills needed by analysts from 5 specific areas:

 Mathematics

Inductive and Deductive Reasoning

Business Fundamentals

Data Concepts

Data Quality

The Skill Set

Analysts need math skills like fish need water, and without them they won’t be analysts, or fish, for very long. There are a set of specific areas within mathematics that are needed for analytical examination. There are obvious ones like understating changes and fluctuations in both relative and absolute terms, peer (competitive) and periodic comparisons and ratios come to mind as the basic skills. Businesses, however, are starting to become more advanced in their analytical view of their data as volume increases and quality improves. Things once reserved for pure statisticians such as acceleration rate of changes, trending and projections, statistical examinations like z-scores, correlations and standard deviations, and data normalization have made their way in to business analytics.

Reasoning seems to becoming a lost art, but it is an important part of analytical study. The ability to observe data and see patterns and from that patters develop a hypothesis and theory of what is occurring and why is important in finding hidden trends or opportunities for growth. Inductive Reasoning takes a curious person to sit down with a large data-set, with no specific question to answer, and parse through it to see what they can find. Deductive Reasoning is a bit more traditional, in cases where you know what is happening, but you need to understand why it is happening.

Business fundamental may seem like an obvious skill, but there are cases where students are coming out of school without understanding business fundamentals from an analytical sense.  Students need a strong understanding of accounting principles like credits and debits, balance sheets items versus income statement items and cost and managerial accounting. Marketing is an important area to understand when it comes to analytics, and understanding areas such as market basket analytics, customer loyalty, and market share versus wallet share as well as how to measure campaign effectiveness are essential. Another area is supply chain management and concepts such as fill rates, lead time and consumption rates.

There is a better than average chance that the last three items I touched on are covered in a good school of business, but basic data concepts are lacking. Without understating concepts like cardinality, parent-child relationships, mutual exclusivity, as well as basic data warehousing fundamentals like facts and dimensions, star schema methodology and table relationships they can be behind the eight-ball on analytical capabilities.

Data quality also falls in to the areas that are not covered in a traditional business school. Without understanding data population, data validity, consistency and completeness of data they may make assumptions on an incomplete or invalid data-set.

The Final Piece; Data Intimacy

The last point to make would be that of data intimacy, which only comes with exposure to the company’s data. An analyst can’t be afraid to dive in, dig through and ask questions so that they can understand every twist, turn, and variation the data may go through or hid in. As an analyst, you have to know the data you are dealing with inside and out.

What does all this mean?

Data volumes are growing at an increased rate, and it is projected that we will not have enough qualified data and business analysts in the very near future based on the demand these data volumes are creating. Higher education must act now to meet the demand of the “real world”, or students must take it on themselves to step outside their schools required curriculum and take class to prepare themselves for the challenges they will face when they enter the workforce. For that to happen, we must educate those students in what the true need is, and not what is prescribed by an outdated curriculum.  By taking the right classes, and preparing themselves from a business and technical sense, students can position themselves to compete in a very high demand job market once they graduate.

Don’t believe that the demand is there? Check out all these articles on The Emerging Role of the Analyst.

Educating the Next Generation of Business Intelligence Professionals

This article, co-written with Christina Rouse, PhD,  discusses educational shortcomings that are occurring as graduates are entering the business intelligence workforce, and we spend time analyzing what changes should occur in the curriculum to address those short comings.
 
Education Article
 
Please visit  the B-Eye Network for the full text.

A Layman’s Guide to CBIP

The Data Warehouse Institute (TDWI) started their CBIP certification offering in 2004 in conjunction with the Institute for Certification of Computing Professionals (ICCP). There are 4 specialty areas that you can focus on with your certification:

Business Analytics
Leadership &Management
Data Analysis & Design
Data Integration

The certification required a series of three tests (minimum) with the first two being the same for all the “tracks” of certifications. The first of the common tests is the IS Core, which tests a broad base of knowledge for general IT/IS laborers. These questions range from needs of a financial reporting system to decoding binary. If you haven’t been in IT organizations for very long you may have a tough time with this test. The second of the common tests is the Data Warehousing Core, which tests general knowledge about data warehousing. Here you will find questions on Kimball and Inmon, full life cycle awareness and general terms and topics as they relate to the practice. The last test is your specialization test, of which you can choose 1 of the 4 listed above, or take all four and get multiple certifications… the choice is yours. I took the Business Analytics test, so that’s the only one I can speak in confidence about. That test focused on OLAP, visualization techniques and data presentation.

I passed my certification tests at the Data Warehouse Institute’s World Conference in Las Vegas in February of 2011. I sat for all three tests during the week, which on top of sitting in the sessions all day, and the fact the conference was in Vegas, made it a challenge in and of itself. Each test lasts a maximum of 90 minutes during which you must answer 110 questions. You also have the option to have the tests administered remotely for an additional fee if you cannot attend the conferences. To actually pass the certification exams you must score a 50% on all three exams (not an average of 50%). If you score 70% or more on all three exams (again, not an average) you will obtain a mastery certification. I scored 61% on my IS Core, 74% on the Data Warehousing Core and 71% on the Business Analytics. I have the option to retake the IS Core to get my mastery, but I don’t think I would fare any better on that test if I were to retake it.

The big question that I get asked is “How did you prepare for the tests?”

Well, I started by ordering the CBIP Preparation Guide from TDWI, but I will tell you that the only thing that it really prepares you for is the testing format because it includes practice exam software. The guide does give you an idea of the types of questions that will be on the test, but as far as strict guidance on what will be on the test; don’t expect that from the guide.

Each sample test in the guide has a section of references, both internet and books. For example, the IS Core exam there is 3 and a half pages of reference materials. I don’t think you would be able to successfully absorb all the information for three tests and even still, the questions being asked will not match anything that appeared in those resources. Truth be told, I never looked at the reference materials and went solely based on my background for a three tests. For the record, I’ve worked in 6 different industries and been doing data warehouse development since 1997.

The other options for preparation are the sessions that are available at the TDWI conferences, I did attend the session on the IS Core and Data Warehousing Core tests, just as a refresher before I took the exams. These sessions are a good indicator of if you should take the tests or not… if you feel comfortable with what is being talked about, you have a better than average change of passing, if you feel like they are talking over your head, just walk away. These tests are designed to test your knowledge that you have gained during the course of your career, and it is not something you can “cram” for and expect to pass.

If you have specific question please email me as I would be happy to answer them as a person who has firsthand experience with the exam process.