Impact of AI-Based Personalized Recommendations on Consumer Purchase Behaviour: A Study Using the Amazon Consumer Behaviour Dataset

Abstract Artificial Intelligence (AI) has transformed digital marketing by enabling e-commerce platforms to analyse consumer behaviour and provide personalized product recommendations. Recommendation systems can reduce information overload, improve product relevance and support customer decision-making. Their effectiveness, however, depends on factors such as recommendation frequency, perceived helpfulness and perceived accuracy. This study examines consumer purchase behaviour using the publicly available Amazon Consumer Behaviour Dataset hosted on Kaggle. The dataset contains 602 observations and 23 variables covering demographics, purchase frequency, browsing behaviour, personalized recommendations, recommendation helpfulness, rating accuracy, reviews, Save-for-Later activity and shopping satisfaction. The study follows a quantitative, explanatory secondary-data design. Reported statistical analyses of the same dataset show significant positive associations between purchase frequency and browsing frequency, Save-for-Later frequency and review helpfulness. A K-means analysis also identifies two distinct behavioural segments. Recommendation-specific analysis reports a positive association between personalized recommendation frequency and rating accuracy. Because the dataset contains Purchase_Frequency rather than a direct Purchase_Intention measure, the findings are interpreted as evidence of consumer purchase behaviour. Keywords: Artificial Intelligence, Personalized Recommendations, Recommendation Systems, Consumer Behaviour, Purchase Frequency, E-commerce, Amazon, Digital Marketing.