Data competition: Predicting the importance of chronic diseases for health institutions

What was this Data Competition about?

• Organizers: ANAP and ATIH (public health French institutions)
• Goal: predict the mid-term impact of chronic diseases for health establishments
• Kind of task: regression
• Metric: RMSE
• Platform: Datascience.net
  • 600 competitors
  • 3 months, ended on 14/10/2016
  • This repository details my solution 9th on both private and public leaderboards

More details can be found at the competition's page

The Data

• Train and Test data with logs of patient visits (age, year, health establishments, number of visits to other health institutions...)
• Information on each health establishment (hospidiag reports)
• All open data you could find and leverage

The data was provided as is, which meant there was still some data engineering needed to work with it

Approach

The detailed approach can be found in French in this ipython notebook