Transcript Career Opportunities in Statistical Computing

Career Opportunities in
Statistical Computing
Robert N. Rodriguez
Director, Statistical Research & Development
[email protected]
Workshop for Chairs of Programs in Statistics
and Biostatistics
August 2, 2008
Copyright © 2006, SAS Institute Inc. All rights reserved.
Two Perspectives on Careers in Statistical Computing
1. Software development opportunities at SAS
2. Emerging opportunities in business
SAS Cary Campus Advanced
Analytics R&D
~100 Ph.D. developers in
statistics, forecasting, data
mining, operations research,
and numerical analysis
Copyright © 2006, SAS Institute Inc. All rights reserved.
SAS Global Reach & Local Presence
Connecting with Customers
 More than 400
offices globally in
51 countries
 10,110 employees
 4.5 million users
worldwide
• Approx. 40,000 sites
• 109 countries
 Hundreds of local
user groups
globally
Copyright © 2006, SAS Institute Inc. All rights reserved.
2007 Worldwide Results
By Industry
Services
11%
Other
2%
Retail
4%
Financial Services
42%
Manufacturing
6%
Healthcare
& Life Sciences
8%
Communications
8%
Government
14%
Energy & Utilities
2%
Copyright © 2006, SAS Institute Inc. All rights reserved.
Education
3%
What’s Involved in Producing Statistical Software?
1. Listening to customers
2. Keeping up with advances in
statistical methodology
3. Designing, writing, testing
code
4. Writing user documentation
5. Providing technical support
6. Consulting with customers
7. Presenting to customers
Copyright © 2006, SAS Institute Inc. All rights reserved.
Statistical software testers
Cheryl LeSaint and Yu Liang
Where Do Statisticians Contribute at SAS?
 Software development
• 30 developers (Ph.D.)
 Software testing
• 20+ testers (M.S. and Ph.D.)
 Documentation
 Technical support
• 15+ statisticians (M.S. and Ph.D.)
 Education
• 12 statisticians (M.S. and Ph.D.)
 Marketing and consulting
Copyright © 2006, SAS Institute Inc. All rights reserved.
Statistical software developers
Randy Tobias and Pushpal
Mukhopadhyay
What Drives Statistical Software Development?
Customer Problems Examples of Development Directions
Complex data
Highly flexible models, Bayesian models,
methods for model selection and validation
Missing data
Multiple imputation
Messy data
Outlier detection, robust methods
Planned data
Survey methods, sample size computation,
design of experiments
Unexplored data
Graphical methods
Large data
Scalable algorithms, parallel processing,
distributed computing
Copyright © 2006, SAS Institute Inc. All rights reserved.
Where Are We Growing?
Mixed modeling
Nonparametric regression
Bayesian modeling
Reliability data analysis
Bayesian econometrics
Statistical graphics
Statistical process control
Survey data analysis
Marketing research methods
Survival data analysis
Data mining, machine learning
Predictive modeling
~20 new specialist positions in development and testing
Copyright © 2006, SAS Institute Inc. All rights reserved.
What We Look for in Statistical Software Developers
 Ph.D. in statistics, biostatistics, applied math, …
• Specialization in one of the target areas
• In-depth knowledge of computational techniques
 Professional programming skills (hard to find!)
• Ability to write large, complex programs in C (not the same as
writing programs in SAS, Matlab, S-PLUS, or R)
• Developed through on-the-job mentoring
 Motivation
• Challenged by creating software that moves new methods into
practice and helps customers solve problems
Copyright © 2006, SAS Institute Inc. All rights reserved.
What We Look for in Statistical Software Testers
 M.S. or Ph.D. in statistics, biostatistics, …
• Graduate coursework in several target areas
• Knowledge of applications and computational methods
 Skills
• Ability to verify computations through validation programs written
in SAS, SAS/IML, SAS macro
• Ability to communicate effectively with other testers and
developers
 Motivation
• Challenged by setting and meeting high standards of accuracy
and performance that exceed customer expectations
Copyright © 2006, SAS Institute Inc. All rights reserved.
Opportunities for Graduate Students
 SAS Summer Fellowship in
Statistical Computing
• opportunity to experience a
professional software development
environment
• competitive; covers stipend, living
expenses
• announced in December Amstat
News
 Permanent Positions
• JSM Placement Service
• Job listings at www.sas.com/jobs
Copyright © 2006, SAS Institute Inc. All rights reserved.
Guixian Lin (2008 Fellow,
U of Illinois) and Robert
Cohen (SAS)
The Business Landscape
Data Flood
Data-Based Decisions
Copyright © 2006, SAS Institute Inc. All rights reserved.
Statistical Computing in Customer Environments
 Data planning
• Design of surveys, experiments, clinical trials, …
 Data access and management
• Disparate data sources and poor data quality undermine analysis
• Databases, data warehouses are controlled by IT (not analysts)
• Statisticians need better skills to participate
 Data preparation
• Getting the data into analysis-ready form (“70% of the effort”)
 Analysis
 Reporting
• Graphics, web pages, FDA submissions, …
Copyright © 2006, SAS Institute Inc. All rights reserved.
Emerging Opportunities in Statistical Computing:
Analytical Software Solutions
• Integrated solutions for key business problems
• Developed by interdisciplinary teams
−
−
−
−
industry experience
software skills (SAS, Java, database, user interface)
statistical computing skills
expertise in formulating and building statistical models
• Examples
−
−
−
−
−
−
monitoring for credit card fraud
credit scoring
customer retention and marketing automation
risk analysis (credit, market, and operational)
web analytics
warranty analysis
Copyright © 2006, SAS Institute Inc. All rights reserved.
Emerging Opportunities in Statistical Computing:
Modeling in Financial and Retail Industries
• Expertise in formulating business problems statistically
• Computational skills in predictive model-building, forecasting,
and optimization
• Examples
− survival models for customer lifetime value
− predictive models for repayment behavior
− forecasting demand for store items
− experimental design for optimizing response in direct marketing
•
Helpful articles for students
− Kahn, “The Practice and Culture of Statistics in Financial Services”,
Sept 2006 Amstat News
− DeVeaux and Ungar, “Careers in Data Mining”, Sept 2007 Amstat News
Copyright © 2006, SAS Institute Inc. All rights reserved.
Emerging Opportunities in Statistical Computing:
Business Analytics
 Enterprise-wide decision-making
based on corporate data using
statistical modeling, forecasting, data
mining, and optimization
 Davenport and Harris (2007)
describe companies that use
analytics to drive performance and
value
 New professional master’s program
in advanced analytics at North
Carolina State University
http://analytics.ncsu.edu
Copyright © 2006, SAS Institute Inc. All rights reserved.
Davenport and Harris (2007),
Competing on Analytics,
Harvard Business School Press