This project aimed to realize a supervised model for 3 problems involving thyroid hormones:
- 1º To assess whether a person has a particular thyroid problem according to their values.
- 2º Trying to guess their gender according to their values.
- 3º Try to guess their sex according to their values.
A dataset with approximately 8000 patient entries was used to train and test the model.
To realize the model, we tested which of the best models had the best result. We applied EDA techniques to the data, selection techniques such as SFS and statistical comparison to identify the best features, and parameter tuning to improve model performance by preventing overfitting and/or underfitting.
- The project was realized in Python using the sklearn library.
- The project was carried out using a Jupyter notebook, so all you have to do to run the project is run the cells.
- Alongside the work is the final report in Portuguese on analyzing the problems with the models made.
- The final evaluation of the project was 19.5/20.