The music industry has undergone enormous change since the introduction of of Napster in 1999. In 1999, 100% of industry revenue was from physical sales; in 2014, United States music industry revenue was 32% physical, 37% digital downloads, 27% streaming, and 4% other minor categories. In this thesis, I present the first models in the music industry that predict monthly revenue at the album level across both revenue stream and geography within the music industry, which are based on a hierarhical linear modeling framework. In addition to the predictive models, I present several sensitivity analyses to examine interesting properties of the data. Specifically, the sensitivity analyses address the effects of data missingness, design imbalance, and the impact of outliers on the predictive results