Personal Data Analysis
case study
Piotr Lewandowski
Last time...
music + computer activity
Rescue time data
- September
- 60 hours logged
- 1000 activities
- avg 17 per hour
- July - October: 4000 records
2: Very productive
1: Productive
0: Neutral
-2: Very unproductive
-1: Unproductive
Last.fm
- July - October
- 2001 listened tracks
- 22 per day
- Prediction: 6000 listened tracks
- Not so much to do Machine Learning
- But... if we measure time
Last.fm problems
- API has more data
- but is more complicated
- Some issues to solve
- API limits
- Not clear data results
Last.fm limits
You will not make more than 5 requests per originating IP address per second, averaged over a 5 minute period, without prior written consent.
Competitors
Relationship between music and productivity
Opinionated correlations
- Track with specified tags
- Not listening music at all
- Gender of music
How to do it better?
Propose music that makes you productive
TODO
- UI for analysing data
- PoC charts
- Multiple users
- Technical
- TypeScript in node.js or Java
- MongoDB -> PostgreSQL
Personal Data Analysis
By Piotr Lewandowski
