Predicting Hourly Internet Traffic in the RTB System – Panel Approach

The aim of this article is to provide a method allowing to control the costs in the real time bidding system related with the bid request traffic. Hourly limits of expenses are predicted based on the historical data. This approach allows to diversify the costs through the whole day, instead spending them immediately at the beginning of the day. To improve the accuracy, a proposed method includes a panel econometric model, where hours are panels. Results are evaluated on the basis of off-line comparison tests between the panel (fixed effects estimator) and non-panel model (ordinary least squares estimator). It turns out, that in most cases the panel method gives more accurate predictions.