Click Through Rate Prediction for Display Advertisement

Abstract

Computational Advertising is the currently emerging multidimensional statistical modeling sub-discipline in digital advertising industry. Web pages visited per user every day is considerably increasing, resulting in an enormous access to display advertisements (ads). The rate at which the ad is clicked by users is termed as the Click Through Rate (CTR) of an advertisement. This metric facilitates the measurement of the effectiveness of an advertisement. The placement of ads in appropriate location leads to the rise in the CTR value that influences the growth of customer access to advertisement resulting in increased profit rate for the ad exchange, publishers and advertisers. Thus it is imperative to predict the CTR metric in order to formulate an efficient ad placement strategy. This paper proposes a predictive model that generates the click through rate based on different dimensions of ad placement for display advertisements using statistical machine learning regression techniques such as multivariate linear regression (LR), poisson regression (PR) and support vector regression(SVR). The experiment result reports that SVR based click model outperforms in predicting CTR through hyperparameter optimization.

Advertising Bidding Optimization by Targeting Based on Self-Learning Database

“…The procedure of the weighting vector evaluation is a nonlinear optimization problem. Different ways of its iterative solution and implementation are discussed [2,5,9,10]. But LR based methods cannot capture higher order interactions between features, which have proved to be important in the CTR prediction [5,7].…”

Click-through rate prediction in online advertising: A literature review

“…In the literature on advertising CTR prediction (e.g., Shan et al, 2016;Deng et al, 2018;Ling et al, 2017;Liao et al, 2014;Li et al, 2015;Zhang et al, 2017), features can be categorized into five classes, as summarized in Table 2: (1) advertising features; (2) user features; (3) context features; (4) query features; and (5) publisher features.…”

DARES: An Asynchronous Distributed Recommender System Using Deep Reinforcement Learning

“…We train the models for all users in all three datasets and their full interaction history. The metric that we use for evaluation is the Click Through Rate (CTR) [5], which is the number of clicks that the users have interacted with an item that was recommended by the model divided by the total number of recommended items.…”