Machine Learning Approach for Analyzing Google Ads Transparency Data

Abstract

In the rapidly evolving world of digital advertising, transparency and accountability have become critical concerns for stakeholders including advertisers, consumers, and regulators. As online advertising strategies become increasingly sophisticated, driven by advanced data analytics and machine learning technologies, ensuring transparency and compliance with advertising regulations is more important than ever. The Google Ads Transparency Center provides a valuable resource for understanding ad practices by offering detailed data on ad placements, targeting criteria, and performance metrics. This thesis investigates the application of machine learning (ML) techniques to the data available in the Google Ads Transparency Center to address these concerns. The study focuses on four primary objectives: (1) analyzing ad delivery patterns and performance metrics to identify trends and insights; (2) assessing the transparency and compliance of advertising practices; (3) developing predictive models to forecast the likelihood of ad removal based on historical data; and (4) providing actionable recommendations to improve ad targeting and compliance. Utilizing BigQuery ML, this research employs clustering algorithms, regression models, and anomaly detection techniques to analyze the transparency center’s datasets. The results reveal significant patterns in ad delivery and performance, insights into compliance issues, and the effectiveness of predictive models in forecasting ad removal. The findings highlight the impact of transparency and compliance on ad effectiveness and offer recommendations for enhancing digital advertising practices. This study contributes to the broader understanding of online advertising by integrating machine learning with transparency data, thus offering valuable insights for advertisers, policymakers, and researchers aiming to improve the effectiveness and integrity of digital advertising.