USING MACHINE LEARNING MODELS TO INVESTIGATE CONSUMER ATTITUDES TOWARD ONLINE BEHAVIORAL ADVERTISING

Abstract

The technique of online behavioral advertising (OBA) is a strategy that has been
widely used in the last decade by businesses and advertisers to deliver targeted advertising messages
to internet users. It is done by utilizing technology to record the habits of online shoppers, including
their searches and the content they visit. Users who browse the internet or use social media view
advertisements relevant to their interests, recent searches, and location. We study Twitter users’
attitudes about targeted ads using five different machine learning models in this research, applying
the CRISP-DM framework. Our primary focus is to develop a benchmark Twitter sentiment dataset
related to targeted ads and implement highly accurate machine learning algorithms to predict tweet
text sentiments when discussing targeted ads. The machine learning algorithms used are Logistic
Regression, Random Forest, Multinomial Naïve Bayes, Multi-Layer Perceptron, and Decision Tree.
We use accuracy, precision, recall, and the F1 measure to evaluate their performance. Logistic Regression using the content-based method provides the utmost accuracy of 0.88. We propose a model
that allows real-time consumer attitude research regarding retargeting ads. The results show that
logistic regression is the most accurate method for predicting customer responses to OBA campaigns
and that retargeting and OBA often cause negative feelings in consumers.