Spotify Advanced SQL Project and Query Optimization

Project Category: Advanced Click Here to get Dataset

Overview

This project involves analyzing a Spotify dataset with various attributes about tracks, albums, and artists using SQL. It covers an end-to-end process of normalizing a denormalized dataset, performing SQL queries of varying complexity (easy, medium, and advanced), and optimizing query performance. The primary goals of the project are to practice advanced SQL skills and generate valuable insights from the dataset.

-- create table
DROP TABLE IF EXISTS spotify;
CREATE TABLE spotify (
    artist VARCHAR(255),
    track VARCHAR(255),
    album VARCHAR(255),
    album_type VARCHAR(50),
    danceability FLOAT,
    energy FLOAT,
    loudness FLOAT,
    speechiness FLOAT,
    acousticness FLOAT,
    instrumentalness FLOAT,
    liveness FLOAT,
    valence FLOAT,
    tempo FLOAT,
    duration_min FLOAT,
    title VARCHAR(255),
    channel VARCHAR(255),
    views FLOAT,
    likes BIGINT,
    comments BIGINT,
    licensed BOOLEAN,
    official_video BOOLEAN,
    stream BIGINT,
    energy_liveness FLOAT,
    most_played_on VARCHAR(50)
);

Project Steps

1. Data Exploration

Before diving into SQL, it’s important to understand the dataset thoroughly. The dataset contains attributes such as:

• Artist: The performer of the track.
• Track: The name of the song.
• Album: The album to which the track belongs.
• Album_type: The type of album (e.g., single or album).
• Various metrics such as danceability, energy, loudness, tempo, and more.

4. Querying the Data

After the data is inserted, various SQL queries can be written to explore and analyze the data. Queries are categorized into easy, medium, and advanced levels to help progressively develop SQL proficiency.

Easy Queries

• Simple data retrieval, filtering, and basic aggregations.

Medium Queries

• More complex queries involving grouping, aggregation functions, and joins.

Advanced Queries

• Nested subqueries, window functions, CTEs, and performance optimization.

5. Query Optimization

In advanced stages, the focus shifts to improving query performance. Some optimization strategies include:

• Indexing: Adding indexes on frequently queried columns.
• Query Execution Plan: Using EXPLAIN ANALYZE to review and refine query performance.

---

15 Practice Questions

Easy Level

1. Retrieve the names of all tracks that have more than 1 billion streams.

   SELECT * FROM spotify 
    WHERE stream >1000000000;

2. List all albums along with their respective artists.

   SELECT distinct album, artist FROM SPOTIFY ORDER BY 1; 
	SELECT distinct album FROM SPOTIFY ORDER BY 1;

3. Get the total number of comments for tracks where licensed = TRUE.

   SELECT SUM(comments) FROM spotify 
   WHERE licensed = 'TRUE';

4. Find all tracks that belong to the album type single.

   SELECT * FROM SPOTIFY 
     WHERE album_type = 'single';
   -- or 
   SELECT * FROM SPOTIFY 
     WHERE album_type ILIKE 'single';

5. Count the total number of tracks by each artist.

SELECT
   artist ,
   COUNT(*) AS Total_no_of_songs
FROM spotify 
GROUP BY artist 
ORDER BY 2 DESC;

Medium Level

6. Calculate the average danceability of tracks in each album.

   SELECT 
     album ,
     AVG(danceability) as avg_danceability
   FROM spotify 
   GROUP BY 1
   ORDER BY 2 desc;
   

7. Find the top 5 tracks with the highest energy values.

   SELECT 
     DISTINCT track,
     MAX(energy)
   FROM spotify 
   GROUP BY track 
   ORDER BY 2 DESC
   LIMIT 5;

8. List all tracks along with their views and likes where official_video = TRUE.

   SELECT 
     track , 
     SUM(views) as total_views,
     SUM(likes) as total_likes
   FROM spotify 
   WHERE official_video = 'true'
   GROUP BY 1
   ORDER BY 2 DESC;
   

9. For each album, calculate the total views of all associated tracks.

   SELECT 
     album , track ,
     SUM(views) as total_views
   FROM spotify
   GROUP BY 1,2;
   

10. Retrieve the track names that have been streamed on Spotify more than YouTube.

    SELECT * from (
    SELECT 
     track , 
     COALESCE(SUM(CASE WHEN most_played_on ='Youtube' THEN stream END),0) as streamed_on_youtube,
     COALESCE(SUM(CASE WHEN most_played_on ='Spotify' THEN stream END),0) as streamed_on_spotify
    FROM spotify 
    GROUP BY track
    ) AS x
    WHERE streamed_on_spotify > streamed_on_youtube;
    

Advanced Level

11. Find the top 3 most-viewed tracks for each artist using window functions.

    /* DENSE_RANK() OVER (
    PARTITION BY column1  -- groups the data 
    ORDER BY column2 DESC) -- Determines the ranking order 
    */ 
    
    WITH ranking_artist
    AS
    (SELECT 
    	artist ,
    	track,
    	SUM(views) total_view,
    	DENSE_RANK() OVER(PARTITION BY artist ORDER BY SUM(views) DESC) as rank
    FROM spotify 
    GROUP BY artist, track
    )
    SELECT * from ranking_artist
    WHERE rank <=3;
    

12. Write a query to find tracks where the liveness score is above the average.

    SELECT 
      track ,
      liveness  -- avg(liveness) = 0.19 for this data
    FROM spotify 
    WHERE liveness > (SELECT AVG(liveness) FROM spotify);
    

13. Use a WITH clause to calculate the difference between the highest and lowest energy values for tracks in each album.

WITH cte
AS
(SELECT 
 album , 
 MAX(energy) as highest_energy,
 MIN(energy) as lowest_energy
FROM spotify 
GROUP BY album
)
SELECT 
 album ,
 highest_energy - lowest_energy as energy_diff
FROM cte;

14. Find tracks where the energy-to-liveness ratio is greater than 1.2.

    SELECT
    	DISTINCT album ,
      track ,
      (energy/liveness) AS energy_to_liveness_ration
    FROM spotify
    WHERE (energy/liveness) > 1.2;
    

15. Calculate the cumulative sum of likes for tracks ordered by the number of views, using window functions.

    SELECT 
     artist,
     track,
     views,
     likes,
     SUM(likes) OVER (ORDER BY views DESC) AS cumulative_likes
    FROM spotify
    ORDER BY views DESC;

Here’s an updated section for your Spotify Advanced SQL Project and Query Optimization README, focusing on the query optimization task you performed. You can include the specific screenshots and graphs as described.

---

Query Optimization Technique

To improve query performance, we carried out the following optimization process:

• Initial Query Performance Analysis Using EXPLAIN

  • We began by analyzing the performance of a query using the EXPLAIN function.
  • The query retrieved tracks based on the artist column, and the performance metrics were as follows:
    • Execution time (E.T.): 7 ms
    • Planning time (P.T.): 0.17 ms
  • Below is the screenshot of the EXPLAIN result before optimization:

• Index Creation on the artist Column

  • To optimize the query performance, we created an index on the artist column. This ensures faster retrieval of rows where the artist is queried.
  • SQL command for creating the index:

    CREATE INDEX idx_artist ON spotify_tracks(artist);

• Performance Analysis After Index Creation

  • After creating the index, we ran the same query again and observed significant improvements in performance:
    • Execution time (E.T.): 0.153 ms
    • Planning time (P.T.): 0.152 ms
  • Below is the screenshot of the EXPLAIN result after index creation:

• Graphical Performance Comparison

  • A graph illustrating the comparison between the initial query execution time and the optimized query execution time after index creation.
  • Graph view shows the significant drop in both execution and planning times:

This optimization shows how indexing can drastically reduce query time, improving the overall performance of our database operations in the Spotify project.

Technology Stack

• Database: PostgreSQL
• SQL Queries: DDL, DML, Aggregations, Joins, Subqueries, Window Functions
• Tools: pgAdmin 4 (or any SQL editor), PostgreSQL (via Homebrew, Docker, or direct installation)

How to Run the Project

1. Install PostgreSQL and pgAdmin (if not already installed).
2. Set up the database schema and tables using the provided normalization structure.
3. Insert the sample data into the respective tables.
4. Execute SQL queries to solve the listed problems.
5. Explore query optimization techniques for large datasets.

---

Next Steps

• Visualize the Data: Use a data visualization tool like Tableau or Power BI to create dashboards based on the query results.
• Expand Dataset: Add more rows to the dataset for broader analysis and scalability testing.
• Advanced Querying: Dive deeper into query optimization and explore the performance of SQL queries on larger datasets.

---

License

This project is licensed under the MIT License.