From 2015-2021, I was a professional classical pianist and opera coach. A large part of my job was to learn operas and play them in rehearsals for opera companies; I would serve as the full orchestra, which would be incorporated at the end of rehearsals and in performances.
Learning an entire opera is demanding and time-consuming. Many are over 2 or 3 hours long, most are in a foreign language (which needs to be learned), and none were written for piano so many of them are extremely difficult to play. Moreover, given that opera has been around since the early 1600s, there are far more operas in the repertoire than would be possible for any one person to learn in an entire lifetime, much less in a few years as a student.
In this project, I explored a dataset of opera performances around the world between 2012-2018 to answer the questions that were top of mind when I was a pianist: what operas should I learn? Who are the most popular composers? What languages/countries produced the most operas? For some composers, would I be able to focus on just one of their pieces, or would I need to learn multiple from some composers? Essentially: what is the most efficient way for me to learn as much of the in-demand repertoire as possible, so I could be as hirable and effective as possible? I knew that some operas were more important and popular than others, but I wanted to quantify that popularity.
This dataset was a comprehensive list of opera performances. The original was found on Kaggle, at this link.
To clean the dataset and ensure its accuracy, I took the following steps:
The main caveat of this dataset is that it only includes up to 2018. Normally this lack of immediate recency wouldn't be relevant; however, the COVID pandemic and social justice conversations in the wake of George Floyd protests have had a resounding impact on the classical music and opera worlds, as everywhere else. I would be interested to see a dataset in a few years from now incorporating these changes to determine any tangible impact on the composers being performed and stories being told in opera.
Another important caveat to keep in mind is that many smaller opera companies, particularly in the United States, would likely be left out of a dataset such as this. It would be very difficult to keep an accurate count of every single operatic performance, especially ones that only have a small audience of dozens. At least in the US, a lot of the most innovative programming of newer and lesser-known composers happens at these small grassroots companies, which wouldn't be reflected in this dataset.
Finally, this dataset of course only has data on operas that have been performed, not that have been written but never performed in this date range. Most pieces by famous composers receive at least a performance or two somewhere in the world, but there's a possibility that not all written works are represented here.
For this project, I used Excel/Google Sheets to clean the dataset and store the results of the queries, SQL to analyze the data and Tableau for visualization.
Links to all relevant items:
The main takeaway is that opera performances are dominated to an extreme degree by power laws. This means that performances are very unevenly distributed: although there are thousands of pieces in existence, a much, much smaller number gets performed with any regularity. The same is true for composers; there are many of them, but a small number dominate the field.
The Pareto Principle states that 80% of outcomes come from 20% of causes, but opera is actually even more extreme in this regard. For example:
Though a lot of composer nationalities were represented (69 total), the top 4 (all European) represent 76% of all performances. Graph
The outputs of many individual composers also follow a power law- for example, Mozart wrote 21 operas, but 90% of his performances came from his top 5. Graph
These statistics may not be good news for the health of the opera industry as a whole, but they do provide some pretty clear takeaways for someone who was in my position:
Most composers maintain similar styles throughout their careers, so by focusing on these few operas by these composers, one could cover an extremely high percentage of all the composers/works that one would be expected to know.
Opera is a difficult medium for composition. A composer has to consider the plot, character development, lyrics, melody, orchestration, and on and on. Given the demands, it's perhaps not surprising that most composers don't write a hit opera on their first try. The most popular opera by a composer who wrote only one is Beethoven's Fidelio, with 748 performances. It's the 36th-most-popular (and Beethoven re-wrote it so many times over the course of many years that it almost can't count as the first try).
Some more statistics to put this in perspective:
The caveat that this dataset doesn't include all written works, only performed works, strengthens this argument. Some composers may have written even more pieces, but they are so forgotten that they received 0 performances in this period. Graph
Opera composition has been a male-dominated field for a long time, for both complex and not-so-complex reasons. But that historical disparity is still relevant in contemporary performances of opera- out of 140,537 performances, only 1,093, or 0.78%, were for pieces composed by women. Graph
Some more worrying numbers:
This would be one of the most interesting trends to observe in recent years, post-2020. Programming more pieces by members of under-represented groups in opera has been a major point of discussion in recent years, more so than it was in the mid-2010s.