Stochastic revenue optimization in online advertising
Agashe, Amod A.
MetadataShow full item record
Online advertising is proving to be the most effective and cost efficient medium of advertising. While there exists plenty of research for advertisers to increase effectiveness of their ad campaign, there is no standard protocol available for web publishers to optimize their revenue. Web publishers need to evaluate their ad space and performance of their advertising contracts to optimize their inventory yield. In this thesis we develop revenue maximization model for web publishers which is formulated as a Stochastic Linear Program with randomness in all component matrices. We review various methods and solution algorithms developed in solving LP with stochastic random variables; followed by developing solution algorithm using Randomized Linear Programming method. In the chapter on numerical experiments we present analysis of variation of model parameters on allocation of page views, which leads to counter-intuitive results. We conclude thesis by putting forward optimal strategies for web publishers and outlining future research work.