Spotify Data Tells Us What Makes a Hit Song
The skills I demoed here can be learned through taking Data Science with Machine Learning bootcamp with NYC Data Science Academy.
Spotify Data Tells Us What Makes a Hit Song
Source: Trusted Reviews
Data Science Introduction
The holy grail of aspiring musical stars of every genre is finding out what makes a mega-hit song. Is it a powerful love ballad or beat thumping club banger? For this exploratory data analysis project, I wanted to take the first steps in finding that grail using data science techniques, the Spotify API, and the Spotify Charts Kaggle Dataset.
Why Spotify you ask? First, Spotify is one of the largest global online music streaming services in the world with 406 million monthly active users, including 180 million paying subscribers, as of December 2021. (Spotify) Second, with the Spotify API, you get access to a wide variety of song characteristics/audio features that provide a quantifiable way to analyze a song.
Furthermore, the Spotify playlist has become the new mixtape tape; something to share your music with friends and family or your public.
Project Goals
- Determine if there is a relationship between song characteristics/audio features and the number of times a hit song is streamed?
- Determine if there is a significant relationship for mega-hit songs in the top 1% of songs that were streamed?
- How does the song genre play into this analysis?
Wrangling the Data
The Kaggle data set had over 26,000,000 entries from multiple markets, a global streaming accumulation, a mix of the top 50 viral songs, and the top 200 daily hits songs. I decided that was too broad and unwieldy I selected the largest market in the data set with the most recent complete year that didn't contain viral songs.
Into the Data...
Since I wanted to find out if there was a relationship between characteristics/audio features and the number of streams it was important to me to figure out what songs were streamed the most.
Spotify Data Tells Us What Makes a Hit Song
It was also important to see which genres were streamed the most.
Song Characteristics/Audio Features
| Structural: | key, mode, time signature, duration, |
| Mood: | Danceability, Valence, Energy, Tempo |
| Context: | Loudness, Speechiness, Instrumentalness |
| Properties: | Liveness, Acousticness |
| Comparative: | Popularity |
Data Analysis
Spotify Data Tells Us What Makes a Hit Song
There wasn’t any significant correlation among hit songs.
Megahits Correlation Characteristics Values
Megahits have slightly more correlation to each other but it is not enough to make any real statements based on this analysis.
Spotify Data Tells Us What Makes a Hit Song
Most streamed Genre: Pop Correlation Characteristics Values
Pop megahits had a strong correlation to valence, a weak correlation to danceability, energy and loudness.
Conclusion
1. There is no magic bullet or holy grail to hit songs. 2. In the US market characteristics of a hit songs widely vary. 3. Megahits have slightly more correlation to each other but it is not enough to make any real statements based on this analysis. 4. As would be expected songs in the same genre have some correlation but I was surprised that based on this analysis it wasn’t significantly greater.
Source: Piqsels
About Author
