Publicação
Click-through rate prediction : a comparative study of ensemble techniques in real-time bidding
| Resumo: | Real-Time Bidding is an automated mechanism to buy and sell ads in real time that uses data collected from internet users, to accurately deliver the right audience to the best-matched advertisers. It goes beyond contextual advertising by motivating the bidding focused on user data and also, it is different from the sponsored search auction where the bid price is associated with keywords. There is extensive literature regarding the classification and prediction of performance metrics such as click-through-rate, impression rate and bidding price. However, there is limited research on the application of advanced machine learning techniques, such as ensemble methods, on predicting click-through rate of real-time bidding campaigns. This paper presents an in-depth analysis of predicting click-through rate in real-time bidding campaigns by comparing the classification results from six traditional classification models (Linear Discriminant Analysis, Logistic Regression, Regularised Regression, Decision trees, k-nearest neighbors and Support Vector Machines) with two popular ensemble learning techniques (Voting and BootStrap Aggregation). The goal of our research is to determine whether ensemble methods can accurately predict click-through rate and compared to standard classifiers. Results showed that ensemble techniques outperformed simple classifiers performance. Moreover, also, highlights the excellent performance of linear algorithms (Linear Discriminant Analysis and Regularized Regression). |
|---|---|
| Autores principais: | Blanc, Maria do Canto e Castro Faria |
| Assunto: | Programmatic Real-Time bidding Click-through-rate Classification Ensemble methods |
| Ano: | 2019 |
| País: | Portugal |
| Tipo de documento: | dissertação de mestrado |
| Tipo de acesso: | acesso aberto |
| Instituição associada: | Universidade Nova de Lisboa |
| Idioma: | inglês |
| Origem: | Repositório Institucional da UNL |
| Resumo: | Real-Time Bidding is an automated mechanism to buy and sell ads in real time that uses data collected from internet users, to accurately deliver the right audience to the best-matched advertisers. It goes beyond contextual advertising by motivating the bidding focused on user data and also, it is different from the sponsored search auction where the bid price is associated with keywords. There is extensive literature regarding the classification and prediction of performance metrics such as click-through-rate, impression rate and bidding price. However, there is limited research on the application of advanced machine learning techniques, such as ensemble methods, on predicting click-through rate of real-time bidding campaigns. This paper presents an in-depth analysis of predicting click-through rate in real-time bidding campaigns by comparing the classification results from six traditional classification models (Linear Discriminant Analysis, Logistic Regression, Regularised Regression, Decision trees, k-nearest neighbors and Support Vector Machines) with two popular ensemble learning techniques (Voting and BootStrap Aggregation). The goal of our research is to determine whether ensemble methods can accurately predict click-through rate and compared to standard classifiers. Results showed that ensemble techniques outperformed simple classifiers performance. Moreover, also, highlights the excellent performance of linear algorithms (Linear Discriminant Analysis and Regularized Regression). |
|---|