Context During some studies developed using Million Song Dataset (https://labrosa.ee.columbia.edu/millionsong/), we needed to clean the subset up to get data in a convinient way to our interests. Content This data consists of the results of cleaning up the MSD subset, available in https://labrosa.ee.columbia.edu/millionsong/pages/getting-dataset#subset. Acknowledgements This is a work developed by Marcos Pedro Ferreira Leal (mleal@ime.usp.br), Shayenne da Luz Moura (shayenne@ime.usp.br) e Thais Rodrigues Neubauer (thais.neubauer@usp.br). Inspiration Some of the questions we want to answer (http://www.cs.colostate.edu/~cs555/CS555-Fall2017-HW3.pdf): Q1. For each artist, what is the most commonly tagged genre? Q2. What is the average tempo across all the songs in the dataset? Q3. What is the median danceability score across all the songs in the dataset? Q4. Who are the top ten artists for fast songs (based on their tempo)? Q5. What are top ten songs based on their hotness in each genre? Please also provide the artist name and title for these songs. Q6. On a per-year basis, what is the mean variance of loudness across the songs within the dataset? Q7. How many songs does each artist have in this dataset? Q8. What are the top ten most popular terms (genres) that songs in the dataset have been tagged with?