Inspiration
The covid crisis has become a turning point in today's society. Although a lot of data has been obtained from patients to perform virus identification techniques, not so much is known about children. That's why we decided to help in identifying the most probable symptoms of covid in child patients.
What it does
After dropping the unsignificant variables and optimizing the correlated variables, we have tried two models of classification: Bayes Networks and Random Forest Classification.
How we built it
We built it in Jupyter Notebooks environment in order to track the data and we used libraries of visualization such as Seaborn (in order to see the heatmap of correlation matrix) and matplotlib. We have use pomegranate and sklearn in order to train the Bayesian Network and Random Forest models respectively.
Challenges we ran into
Most of the difficulties are caused by the lack of complete information.
Accomplishments that we're proud of
At the end, we have acomplished a ~87.9% of well-predictions in Bayesian Networks.
What we learned
We have realized that we can not drop the variables without knowing the relation with the outcome or the dataset, that was one of our biggest errors.
What's next for Predictive Model COVID in Pediatric Age
Integration with FastAPI in order to create an API and a UI.
Log in or sign up for Devpost to join the conversation.