San Francisco Chapter of the American Statistical Association October Seminar



Wenqing Lu, Senior Director, Analytics Department, Orbitz


October 23, 2014  4:30-6:00pm (4:30-5:00pm networking/refreshment; 5-6pm seminar)


San Jose State University (SJSU). Room 202 of the Boccardo Business Complex (BBC)


Methods to Improve Online Search Efficiency




Online search generally follows power law and shows a long tail distribution. With powerful search engine consumers are very specific when searching for their products. Thus the long tail cannot be ignored and in fact largely determines the success of online marketing campaigns. I will look at the long tail problem at Orbitz and how we developed the Bayesian approach and ensemble model to tackle the problem. These methods have advantage over the typical clustering of tail method and can be easily applied to search engine marketing (SEM) and other long tail problems. 


Customer review text contains valuable information that can be used for many purposes. However, it is unstructured data and hard to deal with. With recent development of text mining tools we can effectively turn the unstructured customer review into numerical variables. The new variables can be added to the predictive models along with structured data and significantly improve the model efficiency. I will show how to improve predictive models and gain 40% in online marketing efficiency through text mining hotel reviews at Orbitz Worldwide.


Information for room 202 of the Boccardo Business Complex:

Driving directions to SJSU can be found at  The seminar will take place in room 202 of the Boccardo Business Complex, designated as BBC and located in grid C4 of the campus map at  Attendees driving to the meeting would likely park in either the North Parking Garage (map grid A4) or the South Parking Garage (map grid D2).  One day or partial day parking permits are available inside the parking garages.  Park and then look for the blue and yellow permit machines as shown at Buses and the VTA light rail all have stops near campus.  There is also a shuttle from the local CalTain station directly to SJSU (