Deepak Agarwal

This paper describes, Scout, a statistical modeling driven approach to automatically recommend new Point of Presence (PoP) centers for web sites. PoPs help reduce a website’s page download time dramatically. However, where to build the new PoP centers given the current assets of existing ones is a problem that has rarely been studied in a quantitative and principled way before; it was mainly done through empirical studies or through applying industry experience and intuitions. In this paper, we… Show more

This paper describes, Scout, a statistical modeling driven approach to automatically recommend new Point of Presence (PoP) centers for web sites. PoPs help reduce a website’s page download time dramatically. However, where to build the new PoP centers given the current assets of existing ones is a problem that has rarely been studied in a quantitative and principled way before; it was mainly done through empirical studies or through applying industry experience and intuitions. In this paper, we propose a novel approach that estimates the impact of the PoP centers by building a statistical model using the real user monitoring data collected by the web sites and recommend the next PoPs to build. We also consider the problem of recommending PoPs using other metrics such as user’s number of page views. We show empirically that our approach works well, by experiments that use real data collected from millions of user visits in a major social network site. Show less

This book summarizes research done for a period of 5 years when I was at Yahoo! Research. Much of the work was motivated by our project of personalizing the Yahoo! front page. Many of the methods discussed are deployed both at yahoo! and linkedin.

Visual design plays an important role in online display advertising: changing the layout of an online ad can increase or decrease its effectiveness, measured in terms of click-through rate (CTR) or total revenue. The decision of which layout to use for an ad involves a trade-off: using a layout provides feedback about its effectiveness (exploration), but collecting that feedback requires sacrificing the immediate reward of using a layout we already know is effective (exploitation). To balance… Show more

Visual design plays an important role in online display advertising: changing the layout of an online ad can increase or decrease its effectiveness, measured in terms of click-through rate (CTR) or total revenue. The decision of which layout to use for an ad involves a trade-off: using a layout provides feedback about its effectiveness (exploration), but collecting that feedback requires sacrificing the immediate reward of using a layout we already know is effective (exploitation). To balance exploration with exploitation, we pose automatic layout selection as a contextual bandit problem.

There are many bandit algorithms, each generating a policy which must be evaluated. It is impractical to test each policy on live traffic. However, we have found that offline replay (a.k.a. exploration scavenging) can be adapted to provide an accurate estimator for the performance of ad layout policies at Linkedin, using only historical data about the effectiveness of layouts. We describe the development of our offline replayer, and benchmark a number of common bandit algorithms. Show less

publications