I like to tell people that I’m very fortunate that I chose to do something that I love — analytics — and now what I love to do is something that is very popular in industry. It doesn’t always turn out that way.
I’ve been studying or applying Operations Research (OR) since the mid-1980’s. I love using mathematical tools to help solve a problem and add value to the organization or company that I’m with. My first taste of OR was at the United States Military Academy at West Point, where I graduated with a BS in Mathematics/OR. After serving several years as an artillery officer in the US Army, I left the Army to pursue a career in analytics. Of course, we didn’t call it analytics back then, but that’s the more popular term today. Since earning my MS in OR and PhD in Industrial Engineering (IE) from Georgia Tech, I’ve spent the last 14+ years applying what I’ve learned to large global corporations (Lucent Technologies, UPS, and The Home Depot) and now to a fast-growing start-up (Cardlytics).
One of the great things about being an OR practitioner is that you can apply your craft across a variety of industries. I’ve been able to work in the world’s largest fiber-optic cable plant, for the world’s largest package delivery company, and for the world’s largest home improvement retailer. In each of those, I applied OR techniques to different problems — from manufacturing to fraud detection to revenue management to marketing analytics. Some of my experience applying OR/IE/Analytics to problems in industry includes:
- Optimization to Reduce Setup Times [Lucent Technologies]: Led the operation and improvement of an optimization model that allocated inventory to orders in the most profitable manner. Reduced setup time by 30 hrs/week, reduced handling of fiber spools by 5%, and reduced scrap by 20%, increasing revenue over $5M annually.
- Optimization of Customer Order Fulfillment [Lucent Technologies]: Led development of optimization and simulation models that allocated capacity and resources, identified optimal production strategies, and improved manufacturing process flow. Initiated a process to validate current in-process inventory for a major customer and identify “matched” color sets using CPLEX. Maintained 100% on-time shipments to this customer even under unexpected increased customer demand, while decreasing in-process inventory by 40%.
- Spreadsheet Modeling for Production Planning [Lucent Technologies]: Project leader for fiber production planning using a spreadsheet-based interface to an AMPL/CPLEX optimization model that incorporated hard orders, forecasts, and revenue pricing to create the optimal production plan with respect to end-of-quarter income statements. Resulted in a 55% reduction in raw material inventory.
- Predictive Modeling for Fraud Detection [UPS]: Worked with Corporate Credit to identify risky new accounts opened through UPS.com. Prioritized investigative efforts by impact on revenue. Prevented over $8.5M of fraud in first year.
- Optimization for Empty Asset Balancing [UPS]: Worked with corporate Transportation Finance to develop spreadsheet-based tools that minimized costs, conducted “what if?” scenarios, and determined price points of various movement alternatives. Identified several reduced-cost alternatives with an annual savings of $1.5M in first year.
- Pricing Optimization and Revenue Management [UPS]: Worked with stakeholders in Revenue Management to improve domestic target pricing models and helped develop/implement a market rate-based model for bidding on a portfolio of services to new or existing international customers. The approach improved yield over existing pricing methods and provided a more consistent pricing strategy.
- Suspicious Order Monitoring [UPS]: Worked with UPS Healthcare Logistics to develop a patented approach for identifying orders of controlled substances that are “unusually large or frequent”, as required by federal regulations. In addition to regulatory compliance, the approach increased UPS’s value proposition in the healthcare field and generated at least $1M in new revenue during the first year.
- Predictive Modeling for Market Research/Campaigns [The Home Depot]: Led the development of advanced analytical models for customer response and uplift after receiving direct mail/email communications, as well as “next recommended product” engines for 1-to-1 communications. Incorporated third-party models for customer churn and customer lifetime value into targeting activities. Applied learnings into subsequent models.
- Customer Segmentation Model Enhancement [The Home Depot]: Led the analytics effort to incorporate primary research for the Pro segment into a model enhancement for more precise Pro vs. Consumer segmentation. The enhancement helped speed the adoption of the model by executive leadership and inform the use of the model in all aspects of marketing communications.
Warren E. Hearnes, PhD
A completely unscientific 7067 views and counting…thanks for visiting my blog!