Modeling rare events: online advertisement targeting using machine learning and data mining
published: Dec. 3, 2007, recorded: September 2007, views: 8568
Report a problem or upload filesIf you have found a problem with this lecture or would like to send us extra material, articles, exercises, etc., please use our ticket system to describe your request and upload the data.
Enter your e-mail into the 'Cc' field, and we will keep you updated with your request's status.
The Turn automatic targeting network provides advertisers a revolutionary option for online advertising campaigns (online advertising is a $16 billion industry in 2006 according to the Interactive Advertising Bureau). An advertiser simply inputs its ad into a self-serve console, and Turn does the rest. Unlike many ad networks operating today, Turn incorporates extensive industry expertise and innovative technology from the fields of machine learning, information science and statistics, to truly make online advertising risk-free, relevant, simple, effective, and most importantly, profitable. Unlike traditional ad networks where advertisers need teams of employees to manage manual targeting including selecting sites or selecting and optimizing hundreds of thousands of keywords, the Turn network automatically analyzes and targets ads. Turn’s technology dynamically selects and blends hundreds of variables such as past performance, brand strength, user profiles, action type and site categories to determine the best targets for each ad, thus eliminating guesswork, time and complexity. The Turn network is based on statistical technology that intelligently targets both text and graphical ads. By dynamically and automatically selecting and blending targeting variables, Turn can determine the best ad or group of ads for any situation. Turn offers true pay-forperformance with its unique bidded CPA model. Because advertisers pay for actions that they define, Turn eliminates the risk of worthless or fraudulent clicks. Whether an advertiser is paying for product purchases, site visits, leads, or email signups, the advertiser is in control of what they pay for and when they pay for it. In the context of this problem setting (with billions of ad impressions), this poster will address some key issues in modeling rare events using machine learning and data mining such as uncertainty, the regression versus classification dilemma and feature engineering.
Download slides: mmdss07_shanahan_mre_01.pdf (6.8 MB)
Link this pageWould you like to put a link to this lecture on your homepage?
Go ahead! Copy the HTML snippet !
Write your own review or comment: