Hello. I would like a brief tutorial video that shows how to do the following if possible:
I have a target with 2 variables. Currently, I use the forward feature selection to find the features that will give my model the most accuracy. However, I am only interested in finding the features that best predict one of the target variables. Example:
Target is Type of fruit: Apples or Oranges
With all 50 features, the model is 65% percent accurate.
With the forward feature selection, 5 features were identified that make the model 80% accurate.
However, I don’t care about oranges and am only interested in the features that will most accurately predict if a fruit is an apple.
Can you post a brief tutorial in English that shows how I can do this in KNIME?
You can use any data set you have that has binary target variable. If you know how to do this in Orange, that is also acceptable .