Decoding the Hits: Analyzing the Spotify Top 50 (2019)

Business Problem

With fierce competition in the music industry, data-driven decisions are crucial. Resonance Studios challenged our team to uncover what musical traits define a global hit on Spotify.

Hypotheses:

• Songs with high danceability and energy are more likely to chart globally.
• Big-name artists are overrepresented in the top 50.
• Shorter songs (under 3.5 minutes) perform better on streaming platforms.

We used SQL aggregations and visual analysis to test and support these hypotheses.

---

Dataset

• Source: Kaggle – Top 50 Spotify Songs 2019
• Records: 50
• Features: Track name, artist, genre, energy, valence, BPM, popularity, and more.

---

Data Cleaning & Preparation

Performed in sql-spotify-top50/spotify_proyect_clean-06.ipynb using Python (pandas):

• Renamed columns to snake_case.
• Cleaned track_name of special characters using regex.
• Grouped sub-genres into broader categories for consistency.
• Dropped rows with missing artist_id or genre_id to protect relational integrity.
• Removed duplicates and stripped whitespace from text fields.

---

Database Structure

Implemented in MySQL using project spotify tables.sql, with the analysis queries in sql-spotify-top50/Spotify_Mysql_queries.sql.

Tables:

• artists (artist_id, artist_name)
• genres (genre_id, genre_name)
• track (track_id, track_name, artist_id, genre_id, bpm, energy, valence, etc.)

ERD available in: sql-spotify-top50/RDE.png

Relational integrity enforced via foreign key constraints.

---

Technologies Used

• Python (pandas, seaborn, matplotlib)
• MySQL + Workbench
• Jupyter Notebook
• Git + GitHub
• Google Slides (presentation)

---

Exploratory Data Analysis & Visuals

EDA was conducted using scatterplots, bar charts, histograms, and heatmaps.

• Most tracks fell within high ranges of energy, valence, and danceability.
• Pop and Latin were dominant genres.
• Track durations centered around 3.2–3.5 minutes.

We produced over 4 high-quality visualizations that revealed key genre and feature correlations.

---

SQL Insights

SQL queries performed:

• JOINs between track, artists, and genres
• Aggregates by genre: AVG(valence), AVG(danceability)
• Filtered Top 10 tracks by tempo, valence, and popularity
• GROUP BY queries to analyze genre and artist distribution
• Subqueries to highlight standout combinations

---

Project Structure

.
├── top50.csv                                     # source dataset (Kaggle, Top 50 2019)
├── project spotify tables.sql                    # relational schema (MySQL)
└── sql-spotify-top50/
    ├── spotify_proyect_clean-06.ipynb            # cleaning, EDA and visualisations
    ├── Spotify_Mysql_queries.sql                 # analysis queries
    ├── RDE.png                                   # entity-relationship diagram
    ├── common genres valance danceability.csv    # query output
    ├── most popular tracks energy_danceability.csv
    └── high popular tracks dominant genres based on avarage energy , popularity and danceability.csv

This structure ensures reproducibility and separation of concerns between cleaning, analysis, and database scripts.

---

Presentation Link

📽️ **Google Slides – Final Presentation