Slide 1:1 STATISTICS IN THE CORPORATE WORLD—WHAT I WISHED I HAD KNOWN 50 YEARS AGO! Presentation at 2002 Ohio Statistics Career Day Conference October 25, 2002 Gerry Hahn GE Global Research Center (Retired) Adjunct Professor, RPI gerryhahn@yahoo.com Prepared with significant inputs from Necip Doganaksoy, GE GRC


NEW PH.D.’S COMMENT :2 NEW PH.D.’S COMMENT QUESTION: After one year in industry, what single thing have you found most surprising? ANSWER: How little I know—even about the things I think I know a lot about! Topic of forthcoming book with N. Doganaksoy (Wiley, 2004?) : Statistics in the Corporate World—Filling the Gaps (Tentative Title)


DISCUSSION OUTLINE :3 DISCUSSION OUTLINE Statisticians at GE Some Things Statisticians Do (GE Examples) Six Sigma: An Overview What it Takes to be a Successful Statistician in the Corporate World Some Alternative Career Paths Some Pointers for Success Important Things to Learn in School (and Things I Wished I had Learned) Some Great New Technical Opportunities Academia versus Corporate World: A Comparison Concluding Remarks


STATISTICIANS AT GE: WHO ARE THEY? :4 STATISTICIANS AT GE: WHO ARE THEY? GE Global Research Center (GRC) Applied Statistics Lab (14 statisticians: MS & PhD) Part of GRC Information and Decision Technology including work on operations research/simulation, financial decision-making, web-based marketing, embedded computing, etc. Mission: Guide development of new technology and lead successful applications throughout GE via statistical concepts Individuals within GE’s businesses: Plastics, Aircraft Engines, NBC, GE Capital, etc., mostly MS Plus: All exempt employees receiving instruction in statistics as part of their standard “Six Sigma” training


SOME THINGS STATISTICIANS DO (GE Examples) :5 SOME THINGS STATISTICIANS DO (GE Examples) Develop and transition systems to Help GE Appliances build longer life washing machines by Planning/analyzing accelerated life tests for new designs Statistical process control to ensure high quality is maintained Evaluating field life data to monitor “surprises” Help GE Capital Systems be more profitable by Deciding to whom to give credit cards (and for how much) Determining how to price auto warranty policies Identify delinquent portfolios to acquire for collection Help NBC decide What shows to air on prime time How to schedule these shows How many show promotions to run and when Help GE Aircraft Engines Identify flaws in materials Conduct “just in time” maintenance Assess whether birds can differentiate jet engines from mating calls GOAL: IMPROVE PRODUCTS & OPERATIONS BY DATA-DRIVEN METHODS


SIX SIGMA: AN OVERVIEW :6 SIX SIGMA: AN OVERVIEW Official goal: 3.4 defectives per million opportunities More generally, gain quality improvements using a disciplined, quantitatively based approach Evolution Minimize scrap and rework in manufacturing Design products to meet customer expectations Conduct error-free commercial transactions Reduce variability in delivery times, etc. Help customers succeed Broad company-wide introductory training in statistical tools, design of experiments, simulation, etc GOLDEN AGE OF STATISTICS--BUT NOT NECESSARILY OF STATISTICIANS


WHAT IT TAKES TO BE A SUCCESSFUL STATISTICIANS IN THE CORPORATE WORLD :7 WHAT IT TAKES TO BE A SUCCESSFUL STATISTICIANS IN THE CORPORATE WORLD Passion for real problems Outstanding communications and personal skills Team and leadership skills Strength in mathematics and computer science Knowledge in applications areas Enthusiasm Good listener and situation “sizer-upper” ‘Out of the box” thinker: Proactive Hard worker Flexible and adaptable to change


SOME ALTERNATE CAREER PATHS :8 SOME ALTERNATE CAREER PATHS Statistics expert Statistics manager/leader Applications field professional Business manager Each with steadily increasing levels of responsibility and transition from statistical tools to statistical thinking


Slide 9:9 SOME POINTERS FOR SUCCESS Understanding the Statistician’s Role Regard statistics as way of thinking Use process-oriented approach Recognize problems do not come in textbook form Prepare for change Project Involvement Strive to be team member (rather than “consultant”) Be proactive—lobby for early involvement Focus on problems based on - Importance to business - Likelihood of success - Relevance to general vision


Slide 10:10 SOME NON-TECHNICAL POINTERS (cont’d) Succeeding on the Job Strive to understand “real problem” and big picture (systems thinking) Listen and ask fundamental questions Challenge assumptions (politely) Talk in domain customers’ language Be enthusiastic, positive and involved Search for proactive opportunities Guide others in routine analyses Seek to quantify and demonstrate impact of work


Slide 11:11 IMPORTANT THINGS TO LEARN IN SCHOOL General Learnings Acquire good foundation in application area(s), e.g., chemistry, finance, engineering, biology Get industrial experience (while still in school) Communicate with and present effectively to non-statisticians Understand “user-oriented” (and other) software systems Have some “snappy” stories


Slide 12:12 SOME THINGS I WISHED I HAD LEARNED IN SCHOOL Prior to the data analysis Be passionate about getting right data up front Recognize need for “data cleaning” Understand true variability (versus measurement error) Consider DOE as a sequential learning process General concepts and limitations in data analysis Differentiate analytic and enumerative studies Understand limitations of hypothesis tests Recognize most data are over time Recognize censored data Divide and conquer Use reasonable alternative models/assumptions Start (and end) with graphical analyses


Slide 13:13 SOME THINGS I WISHED I HAD LEARNED IN SCHOOL Analysis tools The use and misuse of the normal distribution The pertinence of percentiles The power of simulation, especially for test planning The most frequent question: “How large a sample do I need?” Confidence, tolerance and prediction intervals—vive la difference! Broader view of statistical process control Course work in - Planning investigations (experiments, surveys, etc) - Time series analysis - Quality assurance - Reliability and product life data analysis - Analysis of massive data sets and statistical computing - Discrete simulation and operations research


Slide 14:14 SOME GREAT NEW TECHNICAL OPPORTUNITY AREAS Product improvement: Continue move from reactive quality control to Up-front design improvement Focus on ensuring high reliability Integration of control theory and SPC Product development: Massive DOE’s for Combinatorial Chemistry, e.g., find “winners” from among 2,916,000 combos of Qualitative formulation variables: 20 catalysts, 20 liquids, 10 active ions Quantitative formulations variables: 3 catalyst amounts, 3 liquid amounts, 3 anion amounts Quantitative process reaction factors: 3 times, 3 temperatures,3 pressures Product service: Just in time maintenance via Cradle to grave reliability assessment Remote monitoring and diagnostics Futuristic: People monitoring Transaction improvement Taming vast data bases for speedy decisions (e.g., credit applications) Reducing variability (e.g., in delivery times) Development of tools that are robust to misuse


Slide 15:15 CONCLUDING REMARKS Sound formal education is necessary, but not sufficient, condition Need understand today’s environment and how to succeed in it Use this understanding to decide whether corporate world is for you Use time in school to get best possible preparation Recognize how little you know—but don’t let it inhibit you For slides (or interest in critiquing forthcoming book and sharing insights), contact gerryhahn@yahoo.com Some recent articles: Hahn, G.J., Deming and the Proactive Statistician, American Statistician, Nov 2002 Hahn, Dognakasoy and Hoerl, The Evolution of Six Sigma, Quality Engineering, 2000 Hahn, Hill, Hoerl and Zinkgraf, Impact of Six Sigma Improvement: A Glimpse into the Future of Statistics, American Statistician, 1999 Hahn and Hoerl, Key Challenges for Statisticians in Business and Industry (with discussion), Technometrics, 1998 P.S. If interested in briefly exploring position at GE CRD, see me during lunch hour or afternoon break