We have trained, tuned and evaluated the following models:
Logistic Regression
Tree based models: Decision Tree, Random Forest, LightGBM, XGBoost, CatBoot
Other models: KNN, SVN
While tree based models are giving a slightly better performance than Logistic Regression, I suggest we use Logistic Regression because it is easy to interpret and would allow the partner to use it to understand feature importances in future.
This issue has been opened to discussion about the team's perspectives on model selection.
We have trained, tuned and evaluated the following models:
While tree based models are giving a slightly better performance than Logistic Regression, I suggest we use Logistic Regression because it is easy to interpret and would allow the partner to use it to understand feature importances in future.
This issue has been opened to discussion about the team's perspectives on model selection.