Skip to content

Latest commit

 

History

History
49 lines (32 loc) · 2.83 KB

README.md

File metadata and controls

49 lines (32 loc) · 2.83 KB
Raw

Spotify Data Analyzer

A project which takes Spotify listening data and creates visualizations of the data.

I made this project to get some practice using the tools I learned in my Data Science Cetification.

The notebooks contained within this project contain more information on my thought process and exploratory analysis as well as some insights I generated into my personal data.

Through my month to month listening, I noticed that the amount of time I listened to Carly Rae Jepsen was far beyond any other artist that I listed to, so I wanted to see how well her features of music represented my listening habits.

The features of Carly's music that were particularly different from my other music is

  • her songs are higher tempo
  • her songs are higher valence (musical positivity)
  • her songs are less acoustic
  • her songs are louder
  • her songs have higher energy
  • her songs are more danceable
  • her songs are more often in a major key

I also created a visual representation of the music in the form of Radar/Spider charts to illustrate the hunch that it is the combination of features that I like more than just the one feature individually. These plots are supposed to be displayed at the end of the analysis_with_api_data.ipynb, but Plotly doesn't play well with GitHub, so below is the only visual.

Here they are:

Carly Rae Jepsen Song Chart Non-Carly Rae Jepsen Song Chart Least Listened Non-Carly Rae Jepsen Song Chart

Upon examining these plots, what do we learn?

  1. I very much value features of loudness, energy, danceability, popularity, high valence, and high tempo
  • Note: Popularity likely has no true bearing on my liking a song i.e. it is correlated with songs that I like, but probably not the cause. This is just an intuitive guess and not necessarily confirmed by the data
  1. I do not value speechiness virtually at all
  2. Carly Rae Jepsen is different from the rest of my other music in that they are less acoustic-y and shorter in length
  • I won't give much weight to the high liveliness oberseved in the CRJ plot as this seems to only come from one song, so it's hardly a significant finding

This project contains a script which will run the artist per month analysis and generate the month by month visualizations.

Example of top artist from every month: top artist every month

The libraries I used are listed in the requirements.txt file.

Running the script is as simple as running this command

python3 PATH_TO_SCRIPT PATH_TO_SPOTIFY_DATA