Event: https://www.blueopen.org/
https://www.blueopen.org/curitiba
- Machine Learning
- IBM Data Science Experience
- Data Science
The project aims to predict child mortality risk based on factors related to the mother and gestation. Statistical techniques such as predictive modelling and machine learning were used, with a high success rate when compared to other consolidated techniques, to predict mortality of newborn children, basing its assumptions on a few selected parameters readily available on current government databases. Only historical data was used, focusing on predictive actions and serving as an extra metric to be used when assessing risk associated with the moment of birth and subsequent year, empowering health professionals and government.
Decision Tree Regressor | Decision Tree Classifier | Random Forest Classifier | Neural Network | |
---|---|---|---|---|
Classificados para o grupo de Risco | 441 | 553 | 433 | 582 |
Obitos classificados corretamente no grupo de Risco (%) | 41% | 55% | 47% | 53% |
Classificados para o grupo de Baixo Risco | 270284 | 270321 | 270362 | 270190 |
Obitos classificados incorretamente no grupo de Baixo Risco (%) | 0.7% | 0.6% | 0.7% | 0.6% |