In the binary classification problem to decide whether it’s a cat, the features like ear shape/face/whiskers are categorical features.
Using a logistic regression model to the above features for prediction will be fruitful or not?
If yes, then why we need decision trees? and how to choose which algorithm(Logistic Regression or decision trees) to pick in such situations?
Hi @Thala,
The default is to use your cv set! With the same training set, train a logistic regression model and a decision trees model, then evaluate them with the same cv set, and see which one does better.
Logistic regression gives you one straight boundary line; decision trees give you more complex boundary lines. With higher complexity in boundary lines, the model is able to fit itself to a more complex dataset!
Cheers,
Raymond