Diesi-stat
Diesi-stat Link to heading
Radio Station Music Statistics with Machine Learning
Overview Link to heading
Diesi-stat is a data visualization and analytics project that analyzes 3 years of radio station play data. Using machine learning (KNN), it categorizes songs and provides insights into listening patterns.
This project was the precursor to the more comprehensive Diesi Analytics platform.
Key Features Link to heading
- 3 Years of Data - Analysis of extensive radio play history
- KNN Classification - Machine learning for song categorization
- Interactive Visualizations - Charts and graphs for data exploration
- Pattern Recognition - Identifies trends in music preferences
Technical Implementation Link to heading
| Component | Technology |
|---|---|
| Frontend | Vue.js |
| ML Model | K-Nearest Neighbors (KNN) |
| Visualization | Interactive charts |
| Data | Radio station play logs |
Machine Learning Approach Link to heading
The KNN model:
- Groups similar songs based on features
- Predicts category for new songs
- Identifies outliers and anomalies
Use Case Link to heading
Understanding what music resonates with listeners helps radio stations:
- Optimize playlists
- Identify emerging trends
- Balance variety and familiarity
Evolution Link to heading
This project evolved into Diesi Analytics, which adds real-time features, a full backend, and more sophisticated ML predictions.