For this problem we executed an approach of "divide and conquer", based on the assumption that generating a competition between the members of the team, and creating a feedback loop from the models created by each member of the team, would generate the best possible model for the 3 problems posed for this competition (Counting, Continuous, and Binary predictions)
For the internal competition we created different workspaces for each person on the team. In there we can see the approach each person took and the corrections made based on feedback from the team.
The folder named "Modelos_finales_y_entregable" contains all selected models and their methodologies.
- Continuous: Modelo_final_continuo.ipynb
- Binary: Modelo_final_binario.ipynb
- Counting: Modelo_conteo.R
The scripts followed for variable selection method of continuous and counting datasets can be found in files "mateo/funciones.R" and "Explicación_Continua_Conteo.Rmd".
The script containing the information for variable selection method of binary dataset can be found in "camilo/Nuevos modelos/Binary.ipynb"