Is random forest better suited than neural network in my scenario?

Question

I am learning Neural network and facing this scenario.

Say in my input, X has 20-30 features, and Y is a classification (e.g. 1,2,3,4, 5).

What I need is to find the features that contribute most to the output Y (i.e. the most important features).

Base on my limited understanding about neural network and machine learning, random forest is the model that can give out this information. For neural network, I am not really sure it can achieve what I want.

Is it possible that someone give more ideas?

Answer 1

I'd use Random Forests for your scenario because RF is well known for determining the importance of different features; the methods are implemented in many Machine Learning frameworks, and it's very simple to understand.

While neural network are powerful, the methods for extracting the importance of individual features are less clear. Neural networks are more appropriate if you are more focused on just results.