Analysis of Billboard Top Songs from 2010 to 2019 using Spotify Data

Online URL: https://cmu-ids-fall-2022-final-project-ids-team-apphome-aormdo.streamlit.app/
Team members:
- Liyan Chang (Contact Person: liyanc@andrew.cmu.edu)
- Haoyu Qi (haoyuq)
- Ran Ju (ranj)
- Vrinda Jindal (vrindaj)
- Tanvi Karandikar (tkarandi)

Link to Paper: https://github.com/CMU-IDS-Fall-2022/final-project-ids-team/blob/main/Report.md

Link to Video: https://drive.google.com/file/d/1WBZXsk8kMoBW9OenC28hTJSlLWpgW4k0/view?usp=sharing

Running Instructions for Software

cd app/
pip install -r requirements.txt
streamlit run Home.py

Abstract

In this project we look at a dataset of the Top 50 Billboard songs from the years 2010 to 2019 and use Spotify's musical features to perform analysis. In particular, we look to answer three overarching questions:

Overall Popularity Analysis: What makes these songs popular?
Trends in Popular Songs: How has popularity changed over time?
Song Recommendations: Can we recommend similar songs from a given input?

Work distribution

Team member	Work done
Liyan	Introduction(report), Recommendation songs for users, edge map
Haoyu	Discussion + Future work(report), data cleaning, information extraction, API parsing
Ran Ju	Related Work(report), edge map, scatter plot, ridge line plot
Vrinda	Methods + Results (report), Trend Analysis (drawing insights), Data Analysis, bubble plot, voilin plots, rec sys design
Tanvi	Methods + Results (report), application design, popularity Analysis (insights), bar plots, pie chart, video creation

Project Phases

The Project was carried out in several well ideated and compartmentalized phases. A somewhat detailed flow of the process can be seen below: