ORC IAP Seminar - "Analytics in Internet Business"
Date: Tuesday, January 31st
Time: 9:30am-3:00pm
Place: 32-155
Schedule:
9:30am - 10:00am - Intro and Continental Breakfast
10:00am - 11:00am - Vivek Farias will discuss High Frequency Revenue Management: Dynamic Allocation With Volatile Demand
11:00am - 12:00pm - Steve Clark will discuss topics of his choosing in the area of email marketing strategies with a top online travel agency
12:00pm - 1:00pm - Lunch (please RSVP by 1/26/12)
1:00pm - 2:00pm - David Krikorian will discuss cloud services and its intersection with business
2:00pm - 3:00pm - Nicole Immorlica will discuss Models of User Search in Sponsored Search Advertising
Student Coordinators: Brian Crimmel, Jane Evans, Virot Ta Chiraphadhanakul
Faculty Coordinator: Retsef Levi
Speaker: Vivek Farias
Title: High Frequency Revenue Management: Dynamic Allocation With Volatile Demand
Bio: Vivek Farias is an Assistant Professor at the MIT Sloan School, affiliated with the Operations Management group and the ORC. Prior to this he was a graduate student at the Information Systems Laboratory in the department of Electrical Engineering at Stanford University. His advisor was Prof. Benjamin Van Roy.
Abstract: We describe a simple, easy to interpret algorithm for a large class of dynamic allocation problems with unknown, volatile demand. Applications include Ad Display problems, network revenue management and revenue management for multi-class processing networks. The method operates in an online fashion and relies on a combination of re-optimization and suitable forecast updates. The method is robust (as witnessed by uniform worst case guarantees for arbitrarily volatile demand) and in the event that demand volatility is not large, the method is simultaneously optimal.
A pilot implementation with data from a mobile ad network suggests that the approach is highly practical. From a theoretical perspective, we introduce a simple new device – a balancing property – that allows us to understand the impact of 'changing bases' in re-optimization; this work provides what we believe are the first uniform performance guarantees for model predictive control.
Speaker: Steve Clark
Bio: Steve Clark is a Senior Consultant with Analytics Operations Engineering—a boutique operations research consulting firm based in Boston, MA—where he has worked extensively on data mining, statistical modeling, and other marketing analytics-related projects. He received his MS in Operations Research from the Massachusetts Institute of Technology in 2001 and his BS in Operations Research from the United States Air Force Academy in 1999. Prior to joining Analytics, Steve served as an officer in the US Air Force, conducting analyses of Air Force maintenance and supply-chain operations. He also served as an Assistant Professor at the US Air Force Academy, teaching Operations Research courses through the Department of Mathematical Sciences.
Abstract: As businesses continue to focus more and more on the strategic importance of customer relationship management (CRM), the ability to execute personalized, customer-specific marketing tactics has become increasingly critical. Online and email marketing vehicles provide a fast, flexible tool to deliver these capabilities to customers. Unfortunately, many companies still struggle with leveraging insights from the vast amounts of data in their data warehouses to turn this strategy into action. This talk will present our experience with a top online travel agency looking to re-vamp their email marketing strategy using insights from data mining and predictive modeling. The discussion will include an overview of the business context, analytical approach, and lessons-learned from this real-world consulting engagement.
Free catered lunch for attendees who RSVP to bcrimmel"at"mit.edu by 1/26/12
Speaker: David Krikorian
Bio: David Krikorian has been running cloud computing systems since 1988: Formerly for about a decade on each MIT's pioneering Athena computing environment and at Akamai Technologies, and in recent years at Google where he works as a Site Reliability Engineer.
Speaker: Nicole Immorlica
Title: Models of User Search in Sponsored Search Advertising
Bio: Nicole Immorlica is an assistant professor of computer science at Northwestern University. She received her B.Sc., M.Eng., and Ph.D. from MIT in computer science. Much of her work is at the intersection of computer science and microeconomics, addressing computational and incentive issues in applied markets like sponsored search auctions, online social networks, and two-sided matching markets. Her work also includes theoretical issues in the areas of mechanism design and game theory. She has received several awards including the Microsoft Faculty Fellowship and the NSF CAREER award.
Abstract: Sponsored search advertising is a booming industry that accounts for a significant fraction of the revenue made by search engine companies. In this market, advertisers bid for slots on search result pages and content pages. The search or content provider then runs an auction to determine which ads to show and what to charge. Ads are displayed alongside content, and advertisers pay the provider when users click on their ads. This talk first gives several theoretical models of this market of increasing complexity, and proceeds to analyze revenue and welfare properties of the resulting equilibria. Our focus is on user search behavior, i.e., how the user scans the ads, and how this impacts the auction. We will see a model of stochastic user search behavior that explains externalities between ads, and a model of utility-maximizing user search behavior which impacts the efficient auction design.
The ORC-sponsored IAP activities page is available at http://student.mit.edu/iap/nsor.html
RSVP by 1/26/12 if you want to join us for lunch. Contact: Brian Crimmel, E40-130, (617) 253-6185, bcrimmel"at"mit.edu
