Improve Accuracy in Advertisement Viewability Prediction using PLC and Logistic Regression

Advertising serves as a crucial driver for promoting products and building their brand presence. In contemporary times, online advertising has gained prominence, becoming a burgeoning industry for both product manufacturers and online publishers. Anticipating and catering to user interests and preferences is of paramount importance when determining the types of advertisements to display on a web page to maximize the number of views and clicks. To address this challenge, our research introduces a hybrid model that combines the probabilistic latent class model and a logistic regression classifier. This model operates in two distinct stages. The Probabilistic Latent Class (PLC) model is employed for the initial stage of advertisement viewability prediction. In existing studies, web page attributes, particularly scroll depth and browser behavior, are utilized for this purpose. However, our proposed approach incorporates additional features such as dwell time, scroll depth level, gender, ad type, and viewed status to enhance viewability prediction using the PLC model. In the second stage, we use a Logistic Regression classifier to categorize the gathered ad view data and determine viewability accuracy. The predicted dataset is retrieved from the database and employed for the classification process via Logistic Regression.