Online Shoppers

该数据集记录了 12,330 次线上购物会话行为，每次会话对应一个独立用户。它包含 18 个特征（其中约 10 个为数值类型，8 个为类别类型），包括用户在不同类别页面上的点击次数及停留时间、跳出率、退出率、页面价值、是否接近特殊日、访问系统/浏览器/地区/流量类型/访客类型是否为回访、新访者、是否周末、月份等信息。每条记录还有一个目标属性Revenue，指示该会话是否最终产生购买。该数据集样本类别不平衡，约 84.5% 的会话未产生购买，剩余约 15.5% 的会话为购买完成。其常用于分类任务，即根据用户行为特征预测其购买意向。

This dataset documents 12,330 online shopping session behaviors, where each session corresponds to an independent user. It encompasses 18 features (roughly 10 numerical and 8 categorical), including the number of clicks and dwell time on category-specific pages, bounce rate, exit rate, page value, proximity to special occasions, access system, browser, region, traffic type, visitor type (whether the visitor is a returning or first-time user), whether the session occurs on a weekend, and the month, etc. Each record also includes a target attribute named Revenue, which indicates whether the session ultimately resulted in a purchase. The dataset exhibits class imbalance: approximately 84.5% of sessions did not result in a purchase, while the remaining ~15.5% of sessions resulted in a completed purchase. It is frequently utilized for classification tasks, specifically to predict user purchase intention based on their behavioral features.