This is a subset of the LFM-1b LastFM dataset (http://www.cp.jku.at/datasets/LFM-1b/), which consists of the listening events (user_events.txt) for three user groups: 1,000 low-mainstream users (low_main_users.txt), 1,000 medium-mainstream users (medium_main_users.txt) and 1,000 high-mainstream users (high_main_users.txt). The mainstreaminess definition used here is the "M_global_R_APC" one from this paper: https://arxiv.org/ftp/arxiv/papers/1912/1912.06933.pdf The format of the three user files is "user , mainstreaminess_value" The format of the user-events file is "user \t artist \t album \t track \t timestamp" Example Python-code for analyzing this dataset as well as more information on the user groups based on mainstreaminess can be found here: https://github.com/domkowald/LFM_processing