Prediction model of optimal bid price based on keyword auction data through machine learning algorithms

Abstract

The RTB system is a bidding system for advertising in a specific area of on-line page. A typical RTB bidding system is a system provided by Google’s search engine. In this paper, we use the data of the Naver advertisement bidding, a representative Korean search engine operated by a private bidding for the RTB system. Especially, in case of online keyword advertisement, the rank can be important factor the online page when a user enters a certain keyword into a search engine. For example, if a search keyword is ranked at the top of an online page, the probability of bid being directly connected will be increased for the link of related keyword. Therefore, the bid price of the keyword is changed according to the rank of the search keyword. In the end, it is necessary to find an appropriate bid price for registering a keyword in a private bid system. In this paper, we propose a prediction modeling mechanism to predict optimal bid price of the keyword in a specific ranking of search engine. In order to predict the optimal bid price and advertising ranking on the online page, we perform feature engineering on the related data set and define the prediction model using the machine learning algorithms for the corresponding data set.