Understanding web user behaviours is essential for optimizing user engagement and platform performance, driving digital transformation, and improving information system design. This study uses web usage mining techniques to analyse user login patterns, role-based activity, browser preferences, and dataset interactions on a dataset hosting platform for Bandung City, Indonesia. By examining two datasets: the Log User Login dataset and the Rekap dataset, we identify trends in login frequency, dataset popularity, and engagement metrics. The problem addressed is the limited understanding of user interactions with online datasets, leading to sub-optimal content recommendations and underutilized datasets. This study bridges this gap by uncovering insights into how users interact with datasets, which datasets are most interesting, and the factors influencing dataset popularity. Our research combines descriptive statistics, correlation analysis, and visualization techniques to reveal behavioural patterns. Findings indicate significant differences in login activity across user roles, the dominance of Chrome and Firefox browsers alongside a notable “Other” category, and the high popularity of datasets related to public infrastructure. Correlation analysis reveals a moderate link between dataset views and downloads, but a weaker connection to shares. These insights can inform the design of improved dataset recommendation systems, enhance user engagement strategies, and refine content delivery mechanisms. Future work should focus on refining dataset categorization, developing role-based personalization, and implementing personalized engagement mechanisms, ultimately contributing to more effective information system management.