Radiohead Mini Data Warehouse

A small ELT pipeline built with Snowflake + dbt + Streamlit using a curated dataset of Radiohead tracks.

1. Extract

Data was collected using the iTunes Search API and saved as: data/itunes_radiohead_raw.csv.

Columns include track name, artist, album, genre, duration (ms), and release date.

2. Load (Snowflake)

The CSV was uploaded into: MALLARD_PROJECT.PUBLIC.ITUNES_RADIOHEAD_RAW.

This is the raw layer.

3. Transform (dbt)

Staging model (stg_radiohead)

Cleans and standardizes fields; derives release_year.

Output: PUBLIC.STG_RADIOHEAD

Analytics models

• album_stats: track count + avg/min/max duration per album
• year_stats: tracks released + avg duration per year

Outputs:

• PUBLIC.ALBUM_STATS
• PUBLIC.YEAR_STATS

4. Dashboard (Streamlit in Snowflake)

An interactive Streamlit app was built using Snowpark.

Features:

• Filters (year range, album)
• KPIs (track count, album count, avg length)
• Charts:
  • Tracks released per year
  • Avg track length per album
• Track explorer (longest/shortest tracks)

5. Project Structure

dw_radiohead/
data/
radiohead_project/ (dbt project)
models/
staging/
analytics/

6. How to Run This Project

1. Create environment

python3 -m venv .venv source .venv/bin/activate pip install -r requirements.txt

2. Load data into Snowflake

-- Run this SQL in Snowflake: CREATE OR REPLACE TABLE ITUNES_RADIOHEAD_RAW AS SELECT * FROM @~/itunes_radiohead_raw.csv;

3. Run dbt models

cd radiohead_project dbt run dbt test

4. Start Streamlit dashboard

streamlit run streamlit_app.py