I'm attempting to do feature selection on the variables on my dataset, but unfortunately the overwhelming majority of them are categorical. Currently they've all been encoded into binary variables, meaning that using something like p values will tell me only whether a specific level of a variable is important - not whether the variable itself is. Is there a recommended or best practice to approach this kind of problem?
Mohammad Mahdi Feraidooni
Dear Rachel
Please see the following paper in feature analysis.
Article A new approach for rotation-invariant and noise-resistant te...
It may be useful for your application.
0 votes
0 thanks
- Badges
- Science topic
More Rachel Bennett's questions See All
Similar questions and discussions