Data is music to his ears. Stephen Brady was just out of college and had landed his dream job. He was working for a country music label in his hometown Nashville, Tennessee. Stephen wasn’t an aspiring musician or music producer; his talents were in data analysis.
A few months after joining Black River Entertainment, Stephen could hardly believe the possibilities before him. He had access to broad and deep data about radio promotion and plays. The potential seemed enormous: He could see how different songs performed as they moved up or down the charts. He could track the viral nature of songs as they spread across geographic areas. And yet, it seemed as if nobody—within the industry, much less his organization—was making use of these data assets.
Stephen set to work building probabilistic models to forecast radio airplay trends. He looked for predictive variables that could help improve the efficiency of radio promotion. His analyses could help the music label promote its artists more effectively by understanding when and where to release singles. He set up a meeting to share his insights with the music label’s CEO Gordon Kerr and the promotional team.
Eighty slides later and after seeing visualizations like the one in Figure 8-1, the promotional team shared a collective look of awe and confusion. What was it supposed to do with his new probabilistic model? What did he mean when he shared a graphic showing "the cosine similarity between ...