Eklavya16/ott-viewer-dropoff-retention

该数据集提供了OTT（流媒体）电视连续剧的剧集级别观看者行为数据，专注于观看者流失模式、保留风险和参与动态。数据集包含450多个独特的电视节目，每个节目最多4季，每季所有剧集都包括。每一行代表一个剧集，支持对季节和剧集进展的细粒度分析。数据来源于TMDB（The Movie Database）API，观看者参与和保留信号是合成生成的，但基于现实的OTT观看行为假设。数据集适用于机器学习、探索性数据分析（EDA）、聚类和时间序列分析等多种用途。

This dataset provides episode-level viewer behavior data for OTT (streaming) TV series, focused on drop-off patterns, retention risk, and engagement dynamics across episodes and seasons. It includes over 450 unique TV shows, with up to 4 seasons per show and all episodes included per season. Each row represents one episode, enabling fine-grained analysis across seasons and episode progressions. The data is sourced from the TMDB (The Movie Database) API, with viewer engagement and retention signals synthetically generated but grounded in realistic OTT viewing behavior assumptions. The dataset is suitable for machine learning, exploratory data analysis (EDA), clustering, and time-series analysis.